Introduction

In this document, we illustrate the use our R package metamicrobiomeR including:

  1. The application of Generalized Additive Model for Location, Scale and Shape (GAMLSS)1 with beta zero inflated (BEZI) family for the analysis of microbiome relative abundance data. The GAMLSS with BEZI family allows examination of microbiome relative abundance data, which ranges from zero to one and is generally zero-inflated. This model also allows adjusting for covariates and can be used for longitudinal or non-longitudinal study design. In addition, the estimates from GAMLSS-BEZI are log(odds ratio) of relative abundances between groups and thus are comparable across studies and thus facilitate straightforward meta-analysis across studies in later stage. We showed some examples to illustrate the performance of the GAMLSS (in comparison with linear/linear mixed effect models (LM) and LM with arcsin squareroot transformation (implemented in MaAsLin software)) using gut microbiome data from the Bangladesh study of Subramanian et al.2 Their data was downloaded from the authors’ website. As additional options to address compositional effects, Geometric Mean of Pairwise Ratios (GMPR) normalization and centered log ratio (CLR) transformation of bacterial taxa composition with different zero-replacement procedures were also implemented.

  2. The application of random effect meta-analysis models for pooling estimates across microbiome studies. This approach allows examination of study-specific effects, heterogeneity across studies and overall effects across studies. We introduced a comprehensive workflow for the analyses of each microbiome study and meta-analysis pooling estimates across microbiome studies. We showed examples for comparison of infant gut microbiome between genders adjusting for breastfeeding status and infant age at stool sample collection in infants <= 6 months. The gut microbiome data used in our examples were from four studies in Bangladesh, Haiti,3 USA(CA_FL),4 and USA(UNC).5

  3. In addition, we implemented the procedures for predicting microbiome age based on relative abundances of bacterial genera using Random Forest model. This was adapted from the original approach proposed by Subramanian et al. We also illustrate the use of linear mixed model (for longitudinal data) or linear model (for non-longitudinal data) for comparison of multiple alpha diversity indexes between groups adjusting for covariates.

Implementation

The metamicrobiomeR package includes the functions below.

Functions Description
taxa.filter Filter relative abundances of bacterial taxa or pathways using prevalence and abundance thresholds
taxa.meansdn Summarize mean, standard deviation of abundances and number of subjects by groups for all bacterial taxa or pathways
taxa.mean.plot Plot mean abundance by groups (from taxa.meansdn output)
taxa.compare Compare relative abundances of bacterial taxa at all levels using GAMLSS or linear/linear mixed effect models (LM) or linear/linear mixed effect models with arcsin squareroot transformation (LMAS)
pathway.compare Compare relative abundances of bacterial functional pathways at all levels using GAMLSS or LM or LMAS. Compare of log(absolute abundances) of bacterial functional pathways at all levels using LM
taxcomtab.show Display the results of relative abundance comparison (from taxa.compare or pathway.compare outputs)
meta.taxa Perform meta-analysis of relative abundance estimates of bacterial taxa or pathways (either from GAMLSS or LM or LMAS) across studies (from combined taxa.compare/pathway.compare outputs of all included studies) using random effect and fixed effect meta-analysis models
metatab.show Display meta-analysis results of bacterial taxa or pathway relative abundances (from meta.taxa output)
meta.niceplot Produce nice combined heatmap and forest plot for meta-analysis results of bacterial taxa and pathway relative abundances (from metatab.show output)
read.multi Read multiple files in a path to R
alpha.compare Calculate average alpha diversity indexes for a specific rarefaction depth, standardize and compare alpha diversity indexes between groups
microbiomeage Predict microbiome age using Random Forest model based on relative abundances of bacterial genera shared with the Bangladesh study

Install ‘metamicrobiomeR’ and other required packages

rm(list=ls()) # clear all
library(devtools)
#install and load package metamicrobiomeR
install_github("nhanhocu/metamicrobiomeR")
library(metamicrobiomeR) 
#Load other needed packages 
library(knitr)
library(plyr)
library(dplyr)
library(gdata)
library(gridExtra)
library(ggplot2)
library(lme4) 
library(lmerTest)
library(mgcv) 
library(meta)

Results

Performance of GAMLSS

GAMLSS is more sensitive than LM or LMAS when there are observed difference

Example 1: Comparison between breastfeeding statuses in infants < 6 months of age

Plot of mean relative abundance by breastfeeding statuses and age at phylum level
data(taxtab.rm7)
taxlist.rm<-taxa.filter(taxtab=taxtab.rm[[5]],percent.filter = 0.05, relabund.filter = 0.00005)
taxa.meansdn.rm<-taxa.meansdn(taxtab=taxtab.rm[[5]],sumvar="bf",groupvar="age.sample")
taxa.meansdn.rm<-taxa.meansdn.rm[taxa.meansdn.rm$bf!="No_BF" &taxa.meansdn.rm$age.sample<=6,]
taxa.meansdn.rm$bf<-drop.levels(taxa.meansdn.rm$bf,reorder=FALSE)
#phylum
p.bf.l2<-taxa.mean.plot(tabmean=taxa.meansdn.rm,tax.lev="l2", comvar="bf", groupvar="age.sample",mean.filter=0.005, show.taxname="short")
p.bf.l2$p

Comparison between breastfeeding statuses adjusting for age of infants at sample collection using GAMLSS
# Comparison of bacterial taxa relative abundance using LMEM or GAMLSS (take some time to run). 
# Note: running time is not long in regular laptop for both analysis (~10s) and meta-analysis (~5s).  
# However, to save time making the tutorial, some saved data/results are loaded for downstream analysis/display.  
#taxacom6.zi.rmg<-taxa.compare(taxtab=taxtab6.rm[[5]],propmed.rel="gamlss",comvar="bf",adjustvar="age.sample",longitudinal="yes",p.adjust.method="fdr")
#load saved results 
data(taxacom6.rmg)
#phylum
kable(taxcomtab.show(taxcomtab=taxacom6.zi.rmg,tax.select=p.bf.l2$taxuse.rm, showvar="bfNon_exclusiveBF", tax.lev="l2",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.bfNon_exclusiveBF ll ul Pr(>|t|).bfNon_exclusiveBF pval.adjust.bfNon_exclusiveBF
5 k__bacteria.p__proteobacteria 0.37 0.11 0.64 0.0053 0.0166
1 k__bacteria.p__actinobacteria -0.37 -0.65 -0.10 0.0083 0.0166
3 k__bacteria.p__firmicutes 0.24 0.00 0.47 0.0468 0.0499
2 k__bacteria.p__bacteroidetes 0.26 0.00 0.53 0.0499 0.0499
Comparison between breastfeeding statuses adjusting for age of infants at sample collection using LM (without transformation)
#taxacom6.rmg<-taxa.compare(taxtab=taxtab6.rm[[5]],propmed.rel="lm",comvar="bf",adjustvar="age.sample",longitudinal="yes",p.adjust.method="fdr")
#phylum
kable(taxcomtab.show(taxcomtab=taxacom6.rmg,tax.select=p.bf.l2$taxuse.rm, showvar="bfNon_exclusiveBF", tax.lev="l2",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.bfNon_exclusiveBF ll ul Pr(>|t|).bfNon_exclusiveBF pval.adjust.bfNon_exclusiveBF
1 k__bacteria.p__actinobacteria -0.11 -0.19 -0.03 0.0066 0.0266
5 k__bacteria.p__proteobacteria 0.06 0.00 0.11 0.0332 0.0665
2 k__bacteria.p__bacteroidetes 0.01 0.00 0.02 0.0580 0.0734
3 k__bacteria.p__firmicutes 0.05 0.00 0.11 0.0734 0.0734
Comparison between breastfeeding statuses adjusting for age of infants at sample collection using LM with arcsin squareroot transformation (LMAS)
#taxacom6.rmg.as<-taxa.compare(taxtab=taxtab6.rm[[5]],propmed.rel="lm",transform="asin.sqrt",comvar="bf",adjustvar="age.sample",longitudinal="yes",p.adjust.method="fdr")
#phylum
kable(taxcomtab.show(taxcomtab=taxacom6.rmg.as,tax.select=p.bf.l2$taxuse.rm, showvar="bfNon_exclusiveBF", tax.lev="l2",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.bfNon_exclusiveBF ll ul Pr(>|t|).bfNon_exclusiveBF pval.adjust.bfNon_exclusiveBF
1 k__bacteria.p__actinobacteria -0.13 -0.23 -0.03 0.0088 0.0207
5 k__bacteria.p__proteobacteria 0.10 0.02 0.17 0.0103 0.0207
2 k__bacteria.p__bacteroidetes 0.03 0.00 0.05 0.0292 0.0390
3 k__bacteria.p__firmicutes 0.07 0.00 0.14 0.0668 0.0668

Example 2: Comparison between infants from 6 months to 2 years of age with solid food introduction after 5 months vs. before 5 months

Mean relative abundance by month of solid food introduction and age at phylum, order and family level
taxa.meansdn.sl5.rm<-taxa.meansdn(taxtab=taxtab.rm[[5]],sumvar="month.food5",groupvar = "age.sample")
taxa.meansdn.sl5.rm<-taxa.meansdn.sl5.rm[taxa.meansdn.sl5.rm$age.sample>6,]
#phylum
p.sl.l2<-taxa.mean.plot(tabmean=taxa.meansdn.sl5.rm,tax.lev="l2", comvar="month.food5", groupvar="age.sample",mean.filter=0.005, show.taxname="short")
p.sl.l2$p

#order
p.sl.l4<-taxa.mean.plot(tabmean=taxa.meansdn.sl5.rm,tax.lev="l4", comvar="month.food5", groupvar="age.sample",mean.filter=0.005, show.taxname="short")
p.sl.l4$p

#family
p.sl.l5<-taxa.mean.plot(tabmean=taxa.meansdn.sl5.rm,tax.lev="l5", comvar="month.food5", groupvar="age.sample",mean.filter=0.005, show.taxname="short")
p.sl.l5$p

Comparison between infants with solid food introduction after 5 months vs. before 5 months adjusting for age of infants at sample collection using GAMLSS
# comparison of bacterial taxa relative abudnance using LMEM or GAMLSS (take some time to run)
#taxacom.6plus.sl5.zi.rmg<-taxa.compare(taxtab=taxtab6plus.rm[[5]],propmed.rel="gamlss",comvar="month.food5",adjustvar="age.sample",longitudinal="yes",p.adjust.method="fdr")
#load saved results
data(taxacom.6plus.sl5.rmg)
#phylum
kable(taxcomtab.show(taxcomtab=taxacom.6plus.sl5.zi.rmg,tax.select=p.sl.l2$taxuse.rm, showvar="food5>5 months", tax.lev="l2",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.month.food5>5 months ll ul Pr(>|t|).month.food5>5 months pval.adjust.food5>5 months
2 k__bacteria.p__bacteroidetes -0.26 -0.42 -0.10 0.0018 0.0070
1 k__bacteria.p__actinobacteria 0.19 0.04 0.34 0.0119 0.0208
3 k__bacteria.p__firmicutes -0.16 -0.30 -0.03 0.0156 0.0208
5 k__bacteria.p__proteobacteria 0.14 -0.02 0.30 0.0861 0.0861 
#order
kable(taxcomtab.show(taxcomtab=taxacom.6plus.sl5.zi.rmg,tax.select=p.sl.l4$taxuse.rm, showvar="food5>5 months", tax.lev="l4",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.month.food5>5 months ll ul Pr(>|t|).month.food5>5 months pval.adjust.food5>5 months
31 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales -0.35 -0.50 -0.21 0.0000 0.0000
26 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales -0.25 -0.42 -0.09 0.0022 0.0076
24 k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales 0.19 0.04 0.34 0.0127 0.0297
29 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales 0.17 0.01 0.32 0.0359 0.0628
39 k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__enterobacteriales 0.16 0.00 0.32 0.0543 0.0760
32 k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales -0.15 -0.31 0.01 0.0662 0.0773
25 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales 0.10 -0.04 0.24 0.1668 0.1668 
#family
kable(taxcomtab.show(taxcomtab=taxacom.6plus.sl5.zi.rmg,tax.select=p.sl.l5$taxuse.rm, showvar="food5>5 months", tax.lev="l5",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.month.food5>5 months ll ul Pr(>|t|).month.food5>5 months pval.adjust.food5>5 months
54 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__prevotellaceae -0.28 -0.45 -0.11 0.0011 0.0107
73 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__ruminococcaceae -0.25 -0.40 -0.09 0.0016 0.0107
74 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__veillonellaceae -0.24 -0.40 -0.08 0.0034 0.0146
65 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__streptococcaceae 0.23 0.07 0.39 0.0049 0.0159
49 k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales.f__bifidobacteriaceae 0.19 0.04 0.34 0.0127 0.0331
62 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__enterococcaceae 0.19 0.02 0.37 0.0321 0.0619
71 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae -0.16 -0.31 -0.01 0.0339 0.0619
69 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__clostridiaceae -0.18 -0.35 -0.01 0.0381 0.0619
85 k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__enterobacteriales.f__enterobacteriaceae 0.16 0.00 0.32 0.0543 0.0784
77 k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales.f__erysipelotrichaceae -0.15 -0.31 0.01 0.0662 0.0861
50 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales.f__coriobacteriaceae 0.10 -0.04 0.24 0.1668 0.1971
52 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__bacteroidaceae -0.01 -0.19 0.17 0.8900 0.9448
63 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__lactobacillaceae -0.01 -0.17 0.15 0.9448 0.9448
Comparison between infants with solid food introduction after 5 months vs. before 5 months adjusting for age of infants at sample collection using LM
#taxacom.6plus.sl5.rmg<-taxa.compare(taxtab=taxtab6plus.rm[[5]],propmed.rel="lm",comvar="month.food5",adjustvar="age.sample",longitudinal="yes",p.adjust.method="fdr")
#phylum
kable(taxcomtab.show(taxcomtab=taxacom.6plus.sl5.rmg,tax.select=p.sl.l2$taxuse.rm, showvar="food5>5 months", tax.lev="l2",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.month.food5>5 months ll ul Pr(>|t|).month.food5>5 months pval.adjust.food5>5 months
2 k__bacteria.p__bacteroidetes -0.03 -0.05 0.00 0.0180 0.0721
3 k__bacteria.p__firmicutes -0.04 -0.11 0.04 0.3253 0.3524
5 k__bacteria.p__proteobacteria 0.01 -0.01 0.04 0.3446 0.3524
1 k__bacteria.p__actinobacteria 0.04 -0.05 0.14 0.3524 0.3524 
#order
kable(taxcomtab.show(taxcomtab=taxacom.6plus.sl5.rmg,tax.select=p.sl.l4$taxuse.rm, showvar="food5>5 months", tax.lev="l4",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.month.food5>5 months ll ul Pr(>|t|).month.food5>5 months pval.adjust.food5>5 months
31 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales -0.06 -0.11 -0.02 0.0088 0.0613
26 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales -0.03 -0.05 0.00 0.0180 0.0631
32 k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales 0.00 -0.01 0.00 0.1989 0.4640
39 k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__enterobacteriales 0.01 -0.01 0.04 0.3020 0.4901
24 k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales 0.04 -0.05 0.13 0.3875 0.4901
29 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales 0.03 -0.04 0.09 0.4201 0.4901
25 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales 0.00 -0.01 0.02 0.5387 0.5387 
#family
kable(taxcomtab.show(taxcomtab=taxacom.6plus.sl5.rmg,tax.select=p.sl.l5$taxuse.rm, showvar="food5>5 months", tax.lev="l5",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.month.food5>5 months ll ul Pr(>|t|).month.food5>5 months pval.adjust.food5>5 months
54 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__prevotellaceae -0.03 -0.05 -0.01 0.0069 0.0898
73 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__ruminococcaceae -0.02 -0.04 0.00 0.1019 0.4487
65 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__streptococcaceae 0.03 -0.01 0.08 0.1484 0.4487
71 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae -0.02 -0.05 0.01 0.1746 0.4487
77 k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales.f__erysipelotrichaceae 0.00 -0.01 0.00 0.1989 0.4487
52 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__bacteroidaceae 0.00 0.00 0.01 0.2327 0.4487
74 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__veillonellaceae -0.01 -0.03 0.01 0.2416 0.4487
85 k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__enterobacteriales.f__enterobacteriaceae 0.01 -0.01 0.04 0.3020 0.4907
49 k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales.f__bifidobacteriaceae 0.04 -0.05 0.13 0.3875 0.5597
69 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__clostridiaceae 0.00 -0.01 0.00 0.5061 0.6367
50 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales.f__coriobacteriaceae 0.00 -0.01 0.02 0.5387 0.6367
62 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__enterococcaceae 0.00 -0.02 0.01 0.7033 0.7607
63 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__lactobacillaceae -0.01 -0.04 0.03 0.7607 0.7607
Comparison between infants with solid food introduction after 5 months vs. before 5 months adjusting for age of infants at sample collection using LM with arcsin squareroot transformation (LMAS)
#taxacom.6plus.sl5.rmg.as<-taxa.compare(taxtab=taxtab6plus.rm[[5]],propmed.rel="lm",transform="asin.sqrt",comvar="month.food5",adjustvar="age.sample",longitudinal="yes",p.adjust.method="fdr")
#phylum
kable(taxcomtab.show(taxcomtab=taxacom.6plus.sl5.rmg.as,tax.select=p.sl.l2$taxuse.rm, showvar="food5>5 months", tax.lev="l2",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.month.food5>5 months ll ul Pr(>|t|).month.food5>5 months pval.adjust.food5>5 months
2 k__bacteria.p__bacteroidetes -0.05 -0.09 -0.01 0.0270 0.1079
5 k__bacteria.p__proteobacteria 0.02 -0.02 0.07 0.2916 0.3451
3 k__bacteria.p__firmicutes -0.04 -0.12 0.04 0.3168 0.3451
1 k__bacteria.p__actinobacteria 0.05 -0.06 0.16 0.3451 0.3451 
#order
kable(taxcomtab.show(taxcomtab=taxacom.6plus.sl5.rmg.as,tax.select=p.sl.l4$taxuse.rm, showvar="food5>5 months", tax.lev="l4",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.month.food5>5 months ll ul Pr(>|t|).month.food5>5 months pval.adjust.food5>5 months
31 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales -0.08 -0.14 -0.02 0.0127 0.0887
26 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales -0.05 -0.09 -0.01 0.0268 0.0939
39 k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__enterobacteriales 0.03 -0.02 0.08 0.2249 0.4308
32 k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales -0.01 -0.04 0.01 0.3621 0.4308
29 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales 0.04 -0.05 0.12 0.3691 0.4308
24 k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales 0.05 -0.06 0.15 0.3693 0.4308
25 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales 0.01 -0.03 0.04 0.6517 0.6517 
#family
kable(taxcomtab.show(taxcomtab=taxacom.6plus.sl5.rmg.as,tax.select=p.sl.l5$taxuse.rm, showvar="food5>5 months", tax.lev="l5",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.month.food5>5 months ll ul Pr(>|t|).month.food5>5 months pval.adjust.food5>5 months
54 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__prevotellaceae -0.06 -0.10 -0.01 0.0106 0.1372
73 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__ruminococcaceae -0.04 -0.08 0.00 0.0652 0.3699
65 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__streptococcaceae 0.05 -0.01 0.12 0.1139 0.3699
69 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__clostridiaceae -0.01 -0.03 0.00 0.1404 0.3699
74 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__veillonellaceae -0.03 -0.07 0.01 0.1423 0.3699
85 k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__enterobacteriales.f__enterobacteriaceae 0.03 -0.02 0.08 0.2249 0.4757
71 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae -0.03 -0.08 0.03 0.2970 0.4757
62 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__enterococcaceae 0.01 -0.02 0.04 0.3343 0.4757
77 k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales.f__erysipelotrichaceae -0.01 -0.04 0.01 0.3621 0.4757
49 k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales.f__bifidobacteriaceae 0.05 -0.06 0.15 0.3693 0.4757
52 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__bacteroidaceae 0.01 -0.01 0.03 0.4025 0.4757
50 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales.f__coriobacteriaceae 0.01 -0.03 0.04 0.6517 0.7060
63 k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__lactobacillaceae -0.01 -0.07 0.06 0.8309 0.8309

GAMLSS detects the difference when there is observed difference and does not detect the difference when there is no observed difference

Example: Mean bacterial taxa relative abundance in infants from 6 months to 2 years of age by diarrhea status and duration of exclusive breastfeeding

taxa.meansdn.dia.exbf2.6plus.rm<-taxa.meansdn(taxtab=taxtab6plus.rm[[5]],sumvar="diarrhea", groupvar="month.exbf2")
#more detail labs
teste<-taxa.meansdn.dia.exbf2.6plus.rm
teste$month.exbf2l<-mapvalues(teste$month.exbf2,from=c("<=2 months",">2 months"),to=c("Duration exbf <=2 months","Duration exbf >2 months"))
#phylum
p.dia.exbf2.6plus.l2<-taxa.mean.plot(tabmean=teste,tax.lev="l2", comvar="diarrhea", groupvar="month.exbf2l",mean.filter=0.005,legend.position="right",ylab="Relative abundance (6 months - 2 years)", show.taxname="short")
p.dia.exbf2.6plus.l2$p

GAMLSS detects the difference when there are observed difference (In infants with duration of exclusive bf <=2 months)

#comparison 

#GAMLSS
#taxacom.6plus.dia.exbf2.zi.rmg<-taxa.compare(taxtab=taxtab6plus.exbf2.rm[[5]],propmed.rel="gamlss",comvar="diarrhea",adjustvar="age.sample",longitudinal="no",p.adjust.method="fdr")
#load saved results 
data(taxacom.dia.exbf2.zi.rmg)
#phylum
kable(taxcomtab.show(taxcomtab=taxacom.6plus.dia.exbf2.zi.rmg,tax.lev="l2",tax.select=p.dia.exbf2.6plus.l2$taxuse.rm,showvar="diarrheaYes",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff = 1))
id Estimate.diarrheaYes ll ul Pr(>|t|).diarrheaYes pval.adjust.diarrheaYes
1 k__bacteria.p__actinobacteria -0.73 -1.12 -0.34 0.0003 0.0011
3 k__bacteria.p__firmicutes 0.49 0.15 0.84 0.0055 0.0109
2 k__bacteria.p__bacteroidetes -0.29 -0.68 0.10 0.1524 0.2032
5 k__bacteria.p__proteobacteria -0.17 -0.54 0.20 0.3729 0.3729 
#LMEM
#taxacom.6plus.dia.exbf2.rmg<-taxa.compare(taxtab=taxtab6plus.exbf2.rm[[5]],propmed.rel="lm",comvar="diarrhea",adjustvar="age.sample",longitudinal="no",p.adjust.method="fdr")
#phylum
kable(taxcomtab.show(taxcomtab=taxacom.6plus.dia.exbf2.rmg,tax.lev="l2",tax.select=p.dia.exbf2.6plus.l2$taxuse.rm,showvar="diarrheaYes",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff=1))
id Estimate.diarrheaYes ll ul Pr(>|t|).diarrheaYes pval.adjust.diarrheaYes
3 k__bacteria.p__firmicutes 0.08 0.01 0.16 0.0329 0.1317
1 k__bacteria.p__actinobacteria -0.07 -0.15 0.02 0.1304 0.2608
2 k__bacteria.p__bacteroidetes -0.02 -0.05 0.02 0.3068 0.4091
5 k__bacteria.p__proteobacteria 0.01 -0.04 0.06 0.5909 0.5909 
#LMEM with arcsin squareroot transformation
#taxacom.6plus.dia.exbf2.rmg.as<-taxa.compare(taxtab=taxtab6plus.exbf2.rm[[5]],propmed.rel="lm",transform="asin.sqrt",comvar="diarrhea",adjustvar="age.sample",longitudinal="no",p.adjust.method="fdr")
kable(taxcomtab.show(taxcomtab=taxacom.6plus.dia.exbf2.rmg.as,tax.lev="l2",tax.select=p.dia.exbf2.6plus.l2$taxuse.rm,showvar="diarrheaYes",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff=1))
id Estimate.diarrheaYes ll ul Pr(>|t|).diarrheaYes pval.adjust.diarrheaYes
3 k__bacteria.p__firmicutes 0.11 0.01 0.20 0.0269 0.0848
1 k__bacteria.p__actinobacteria -0.12 -0.23 0.00 0.0424 0.0848
2 k__bacteria.p__bacteroidetes -0.06 -0.12 0.01 0.0852 0.1136
5 k__bacteria.p__proteobacteria 0.00 -0.07 0.08 0.9060 0.9060

GAMLSS does not detect the difference when there are no observed difference (In infants with duration of exclusive bf >2 months)

#GAMLSS
#taxacom.6plus.dia.exbf2plus.zi.rmg<-taxa.compare(taxtab=taxtab6plus.exbf2plus.rm[[5]],propmed.rel="gamlss",comvar="diarrhea",adjustvar="age.sample",longitudinal="no",p.adjust.method="fdr")
#phylum
kable(taxcomtab.show(taxcomtab=taxacom.6plus.dia.exbf2plus.zi.rmg,tax.lev="l2",tax.select=p.dia.exbf2.6plus.l2$taxuse.rm,showvar="diarrheaYes",readjust.p=TRUE,p.adjust.method="fdr",p.cutoff=1))
id Estimate.diarrheaYes ll ul Pr(>|t|).diarrheaYes pval.adjust.diarrheaYes
6 k__bacteria.p__proteobacteria 0.12 -0.33 0.56 0.6043 0.9243
2 k__bacteria.p__bacteroidetes 0.07 -0.41 0.56 0.7680 0.9243
4 k__bacteria.p__firmicutes -0.02 -0.40 0.36 0.9142 0.9243
1 k__bacteria.p__actinobacteria 0.02 -0.42 0.46 0.9243 0.9243 
#LMEM
#taxacom.6plus.dia.exbf2plus.rmg<-taxa.compare(taxtab=taxtab6plus.exbf2plus.rm[[5]],propmed.rel="lm",comvar="diarrhea",adjustvar="age.sample",longitudinal="no",p.adjust.method="fdr")
#phylum
kable(taxcomtab.show(taxcomtab=taxacom.6plus.dia.exbf2plus.rmg,tax.lev="l2",tax.select=p.dia.exbf2.6plus.l2$taxuse.rm,showvar="diarrheaYes",p.adjust.method="fdr",p.cutoff=1))
id Estimate.diarrheaYes ll ul Pr(>|t|).diarrheaYes pval.adjust.diarrheaYes
1 k__bacteria.p__actinobacteria 0.02 -0.08 0.12 0.6956 0.9879
6 k__bacteria.p__proteobacteria -0.01 -0.06 0.05 0.7433 0.9879
2 k__bacteria.p__bacteroidetes 0.01 -0.04 0.05 0.8051 0.9879
4 k__bacteria.p__firmicutes -0.01 -0.10 0.08 0.8230 0.9879 
#LMEM with arcsin squareroot transformation
#taxacom.6plus.dia.exbf2plus.rmg.as<-taxa.compare(taxtab=taxtab6plus.exbf2plus.rm[[5]],propmed.rel="lm",transform="asin.sqrt",comvar="diarrhea",adjustvar="age.sample",longitudinal="no",p.adjust.method="fdr")
#phylum
kable(taxcomtab.show(taxcomtab=taxacom.6plus.dia.exbf2plus.rmg.as,tax.lev="l2",tax.select=p.dia.exbf2.6plus.l2$taxuse.rm,showvar="diarrheaYes",p.adjust.method="fdr",p.cutoff=1))
id Estimate.diarrheaYes ll ul Pr(>|t|).diarrheaYes pval.adjust.diarrheaYes
6 k__bacteria.p__proteobacteria 0.02 -0.06 0.11 0.5875 0.9191
2 k__bacteria.p__bacteroidetes 0.01 -0.07 0.09 0.8101 0.9707
1 k__bacteria.p__actinobacteria -0.01 -0.13 0.12 0.8927 0.9707
4 k__bacteria.p__firmicutes 0.00 -0.10 0.10 0.9626 0.9989

Illustration of meta-analysis workflow

This section illustrate the workflow and examples for the analysis of one microbome study comparing gut microbiome between male vs. female infants <=6 months of age adjusting for feeding status and age at stool sample collection and then meta-analysis across four studies.

Summary of four included studies

#Load example Bangladesh study 
data(sam.rm)
patht<-system.file("extdata/QIIME_outputs/Bangladesh/tax_mapping7", package = "metamicrobiomeR", mustWork = TRUE)
taxrel.rm<-read.multi(patht=patht,patternt=".txt",assignt="no",study="Subramanian et al 2014 (Bangladesh)")
# Bangladesh healthy cohort only and add other meta data
taxrel.ba<-list()
for (j in 1:length(taxrel.rm)){
  taxrel.ba[[j]]<-merge(merge(taxrel.rm[[j]][taxrel.rm[[j]]$x.sampleid %in% samde$fecal.sample.id,],samde, by.x="x.sampleid",by.y="fecal.sample.id"), he50[,c("child.id","gender","month.exbf","month.food")],by.x="personid", by.y="child.id")
}
#all samples from birth to 2 years of age 
nrow(taxrel.ba[[1]])
[1] 995
# number of samples for all four studies (for samples <=6 months of age) 
data(studysum)
kable(studysum)
female male all
Subramanian et al 2014 (Bangladesh) 180 142 322
Bender et al 2016 (Haiti) 25 21 46
Pannaraj et al 2017 (USA(CA_FL)) 120 101 221
Thompson et al 2015 (USA(NC)) 14 7 21
Sum 339 271 610

Comparison of bacterial taxa relative abundance between male vs. female infants < 6 months adjusting for breastfeeding statuses and age of infants at sample collection with GAMLSS

Analysis of Bangladesh data

# Comparison of bacterial taxa relative abundance up to genus level (take some time to run)
#taxacom6.zi.rm.sex.adjustbfage<-taxa.compare(taxtab=taxtab6.rm[[5]],propmed.rel="gamlss",comvar="gender",adjustvar=c("bf","age.sample"),longitudinal="yes")
# load saved results 
data(taxacom.rm.sex.adjustbfage)
#phylum
kable(taxcomtab.show(taxcomtab=taxacom6.zi.rm.sex.adjustbfage,tax.select="none", showvar="genderMale", tax.lev="l2",p.adjust.method="fdr"))

id Estimate.genderMale ll ul Pr(>|t|).genderMale pval.adjust.genderMale — ——————– — — ——————– ———————–

#order
kable(taxcomtab.show(taxcomtab=taxacom6.zi.rm.sex.adjustbfage,tax.select="none", showvar="genderMale", tax.lev="l4",p.adjust.method="fdr"))
id Estimate.genderMale ll ul Pr(>|t|).genderMale pval.adjust.genderMale
18 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales -0.34 -0.57 -0.11 0.0040 0.1249
17 k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales 0.28 0.05 0.52 0.0204 0.3061 
#family
kable(taxcomtab.show(taxcomtab=taxacom6.zi.rm.sex.adjustbfage,tax.select="none", showvar="genderMale", tax.lev="l5",p.adjust.method="fdr"))
id Estimate.genderMale ll ul Pr(>|t|).genderMale pval.adjust.genderMale
35 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales.f__coriobacteriaceae -0.34 -0.57 -0.11 0.0040 0.1249
34 k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales.f__bifidobacteriaceae 0.28 0.05 0.52 0.0204 0.3061
36 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__bacteroidaceae -0.29 -0.58 0.00 0.0491 0.4453
51 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__eubacteriaceae -0.98 -1.95 -0.01 0.0493 0.4453 
#genus
kable(taxcomtab.show(taxcomtab=taxacom6.zi.rm.sex.adjustbfage,tax.select="none", showvar="genderMale", tax.lev="l6",p.adjust.method="fdr"))
id Estimate.genderMale ll ul Pr(>|t|).genderMale pval.adjust.genderMale
114 k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales.f__erysipelotrichaceae.g__.eubacterium. -0.70 -1.15 -0.25 0.0024 0.1249
72 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales.f__coriobacteriaceae.g__collinsella -0.35 -0.63 -0.06 0.0165 0.3061
69 k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales.f__bifidobacteriaceae.g__bifidobacterium 0.28 0.05 0.52 0.0204 0.3061
99 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae.g__.ruminococcus. 0.42 0.06 0.79 0.0222 0.3061
74 k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__bacteroidaceae.g__bacteroides -0.29 -0.58 0.00 0.0491 0.4453
93 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__eubacteriaceae.g__pseudoramibacter_eubacterium -0.98 -1.95 -0.01 0.0493 0.4453

The analysis for other studies was done similarly.

Meta-analysis of four studies (Bangladesh, Haiti, USA(CA_FL), USA(NC))

Heatmap of log(odds ratio) (log(OR)) of relative abundances of gut bacterial taxa at different taxonomic levels between male vs. female infants for each study and pooled estimates (meta-analysis) across all studies with 95% confidence intervals (95% CI) (forest plot). All log(OR) estimates of each bacterial taxa from each study were from Generalized Additive Models for Location Scale and Shape (GAMLSS) with beta zero inflated family (BEZI) and were adjusted for feeding status and age of infants at sample collection. Pooled log(OR) estimates and 95% CI (forest plot) were from random effect meta-analysis models with inverse variance weighting and DerSimonian-Laird estimator for between-study variance based on the adjusted log(OR) estimates and corresponding standard errors of all included studies. Bacterial taxa with p-values for differential relative abundances <0.05 were denoted with * and those with p-values <0.0001 were denoted with **. Pooled log(OR) estimates with pooled p-values<0.05 are in red and those with false discovery rate (FDR) adjusted pooled p-values <0.1 are in triangle shape. Missing (unavailable) values are in white. USA: United States of America; CA: California; FL: Florida; NC: North Carolina.

# load saved results of four studies for the comparison of bacterial taxa relative abundance between genders adjusted for breastfeeding and infant age at sample collection 
data(taxacom.rm.sex.adjustbfage)
data(taxacom.ha.sex.adjustbfage)
data(taxacom6.zi.usbmk.sex.adjustbfage)
data(taxacom6.unc.sex.adjustedbfage)
taxacom6.zi.rm.sex.adjustbfage$study<-"Subramanian et al 2014 (Bangladesh)"
taxacom6.zi.rm.sex.adjustbfage$pop<-"Bangladesh"
taxacom.zi.ha.sex.adjustbfage$study<-"Bender et al 2016 (Haiti)"
taxacom.zi.ha.sex.adjustbfage$pop<-"Haiti"
taxacom6.zi.usbmk.sex.adjustbfage$study<-"Pannaraj et al 2017 (USA(CA_FL))"
taxacom6.zi.usbmk.sex.adjustbfage$pop<-"USA(CA_FL)"
taxacom6.zi.unc.sex.adjustedbfage$study<-"Thompson et al 2015 (USA(NC))"
taxacom6.zi.unc.sex.adjustedbfage$pop<-"USA(NC)"
tabsex4<-rbind.fill(taxacom6.zi.rm.sex.adjustbfage,taxacom.zi.ha.sex.adjustbfage,taxacom6.zi.usbmk.sex.adjustbfage,taxacom6.zi.unc.sex.adjustedbfage)
# meta-analysis (take some time to run)
#metab.sex<-meta.taxa(taxcomdat=tabsex4,summary.measure="RR",pool.var="id",studylab="study",backtransform=FALSE,percent.meta=0.5,p.adjust.method="fdr")
#load saved results 
data(metab.sex)
#phylum
kable(metatab.show(metatab=metab.sex$random,com.pooled.tab=tabsex4,tax.lev="l2",showvar="genderMale",p.cutoff.type="p", p.cutoff=0.05,display="table"))

id estimate ll ul p p.adjust — ——— — — — ———

#plot
metadat<-metatab.show(metatab=metab.sex$random,com.pooled.tab=tabsex4,tax.lev="l2",showvar="genderMale",p.cutoff.type="p", p.cutoff=1,display="data")
meta.niceplot(metadat=metadat,sumtype="taxa",level="main",p="p",p.adjust="p.adjust",phyla.col="rainbow",p.sig.heat="yes",heat.forest.width.ratio =c(1.5,1),leg.key.size=0.8,leg.text.size=10,heat.text.x.size=10,heat.text.x.angle=0,forest.axis.text.y=8,forest.axis.text.x=10, point.ratio = c(4,2),line.ratio = c(2,1))

#order 
kable(metatab.show(metatab=metab.sex$random,com.pooled.tab=tabsex4,tax.lev="l4",showvar="genderMale",p.cutoff.type="p", p.cutoff=0.05,display="table"))
id estimate ll ul p p.adjust
18 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales -0.26 -0.44 -0.08 0.0049 0.209 
#some different plot options: increase size of forest plot vs. heatmap, change color palette, legend size 
metadat<-metatab.show(metatab=metab.sex$random,com.pooled.tab=tabsex4,tax.lev="l4",showvar="genderMale",p.cutoff.type="p", p.cutoff=1,display="data")
meta.niceplot(metadat=metadat,sumtype="taxa",level="sub",p="p",p.adjust="p.adjust",phyla.col="rainbow",leg.key.size=1,leg.text.size=8,heat.text.x.size=6,forest.axis.text.y=8,forest.axis.text.x=6,heat.forest.width.ratio =c(1,1.3), neg.palette = "Greens",pos.palette = "Purples", point.ratio = c(4,2),line.ratio = c(2,1))

# family 
kable(metatab.show(metatab=metab.sex$random,com.pooled.tab=tabsex4,tax.lev="l5",showvar="genderMale",p.cutoff.type="p", p.cutoff=0.05,display="table"))
id estimate ll ul p p.adjust
35 k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales.f__coriobacteriaceae -0.26 -0.44 -0.08 0.0049 0.2090
50 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__eubacteriaceae -0.68 -1.35 -0.02 0.0436 0.9833 
#(not show significant p-values of each study in heatmap)
metadat<-metatab.show(metatab=metab.sex$random,com.pooled.tab=tabsex4,tax.lev="l5",showvar="genderMale",p.cutoff.type="p", p.cutoff=1,display="data")
meta.niceplot(metadat=metadat,sumtype="taxa",level="sub",p="p",p.adjust="p.adjust",phyla.col="rainbow",leg.key.size=1,p.sig.heat ="no",leg.text.size=8,heat.text.x.size=7,forest.axis.text.y=8,forest.axis.text.x=7, point.ratio = c(4,2),line.ratio = c(2,1))

#genus 
kable(metatab.show(metatab=metab.sex$random,com.pooled.tab=tabsex4,tax.lev="l6",showvar="genderMale",p.cutoff.type="p", p.cutoff=0.05,display="table"))
id estimate ll ul p p.adjust
165 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae.g__coprococcus 0.43 0.42 0.44 0.0000 0.0000
108 k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales.f__erysipelotrichaceae.g__.eubacterium. -0.44 -0.76 -0.13 0.0056 0.2090
100 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__veillonellaceae.g__megamonas -0.47 -0.91 -0.02 0.0410 0.9833
87 k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__eubacteriaceae.g__pseudoramibacter_eubacterium -0.68 -1.35 -0.02 0.0436 0.9833 
#some different plot options: pooled estimates in forest plot with the same color scales as heatmap, those with p-values<0.05 in bold, FDR adjusted p-values<0.1 in triangles
metadat<-metatab.show(metatab=metab.sex$random,com.pooled.tab=tabsex4,tax.lev="l6",showvar="genderMale",p.cutoff.type="p", p.cutoff=1,display="data")
meta.niceplot(metadat=metadat,sumtype="taxa",level="sub",p="p",p.adjust="p.adjust",phyla.col="rainbow",p.sig.heat="yes",heat.forest.width.ratio =c(1,1.3),forest.col="by.estimate",leg.key.size=0.8,leg.text.size=10,heat.text.x.size=6,forest.axis.text.y=7,forest.axis.text.x=6, point.ratio = c(4,2),line.ratio = c(2,1))

Comparison of relative abundance of bacterial functional (KEGG) pathways between male vs. female infants < 6 months adjusting for breastfeeding statuses and age of infants at sample collection

Analysis for Bangladesh study

# RM
data(sam.rm)
patht<-system.file("extdata/QIIME_outputs/Bangladesh/picrust", package = "metamicrobiomeR", mustWork = TRUE)
kegg<-read.multi(patht=patht,patternt=".txt",assignt="no")
kegg.rm<-list()
for (i in 1:length(kegg)){
  rownames(kegg[[i]])<-kegg[[i]][,"kegg_pathways"]
  kegg[[i]]<-kegg[[i]][,colnames(kegg[[i]])[!colnames(kegg[[i]]) %in% c("otu.id","kegg_pathways")]]
  kegg.rm[[i]]<-as.data.frame(t(kegg[[i]]))
}
covar.rm<-merge(samde, he50[,c("child.id","gender","zygosity","day.firstsample","day.lastsample","n.sample","sampling.interval.msd","month.exbf","month.food",
                               "n.diarrhea.yr","percent.time.diarrhea","fraction.antibiotic","subject.allocation")], by="child.id")
covar.rm<-dplyr::rename(covar.rm,sampleid=fecal.sample.id, personid=child.id ,age.sample=age.months)
covar.rm$bf<-factor(covar.rm$bf, levels=c('ExclusiveBF','Non_exclusiveBF','No_BF'))

covar.rm$personid<-as.factor(covar.rm$personid)
# Comparison of pathway relative abundances (take time to run)
#pathcom.rm6.rel.gamlss.sexg<-pathway.compare(pathtab=kegg.rm,mapfile=covar.rm,sampleid="sampleid",pathsum="rel",stat.med="gamlss",comvar="gender",adjustvar=c("age.sample","bf"),longitudinal="yes",p.adjust.method="fdr",percent.filter=0.05,relabund.filter=0.00005,age.limit=6)
# load saved results 
data(pathcom.rm6.rel.gamlss.sexg)
kable(taxcomtab.show(taxcomtab=pathcom.rm6.rel.gamlss.sexg$l2, sumvar="path",tax.lev="l2",tax.select="none",showvar="genderMale", p.adjust.method="fdr",p.cutoff=0.05))
id Estimate.genderMale ll ul Pr(>|t|).genderMale pval.adjust.genderMale
1 Metabolism..Amino.Acid.Metabolism 0.02 0.00 0.04 0.0302 0.2739
14 Genetic.Information.Processing..Folding..Sorting.and.Degradation -0.01 -0.01 0.00 0.0331 0.2739
9 Organismal.Systems..Endocrine.System 0.07 0.00 0.13 0.0391 0.2739
19 Human.Diseases..Infectious.Diseases -0.03 -0.06 0.00 0.0404 0.2739
2 Metabolism..Biosynthesis.of.Other.Secondary.Metabolites 0.03 0.00 0.06 0.0413 0.2739 
kable(taxcomtab.show(taxcomtab=pathcom.rm6.rel.gamlss.sexg$l3, sumvar="path",tax.lev="l3",tax.select="none",showvar="genderMale", p.adjust.method="fdr",p.cutoff=0.05))
id Estimate.genderMale ll ul Pr(>|t|).genderMale pval.adjust.genderMale
15 Metabolism..Amino.Acid.Metabolism..Arginine.and.proline.metabolism 0.02 0.01 0.04 0.0055 0.565
44 Genetic.Information.Processing..Folding..Sorting.and.Degradation..Chaperones.and.folding.catalysts -0.02 -0.04 -0.01 0.0085 0.565
139 Metabolism..Amino.Acid.Metabolism..Phenylalanine..tyrosine.and.tryptophan.biosynthesis 0.05 0.01 0.08 0.0112 0.565
140 Metabolism..Biosynthesis.of.Other.Secondary.Metabolites..Phenylpropanoid.biosynthesis 0.09 0.02 0.16 0.0159 0.565
33 Metabolism..Carbohydrate.Metabolism..C5.Branched.dibasic.acid.metabolism 0.05 0.01 0.08 0.0168 0.565
38 Metabolism..Energy.Metabolism..Carbon.fixation.pathways.in.prokaryotes -0.02 -0.03 0.00 0.0251 0.565
123 Metabolism..Glycan.Biosynthesis.and.Metabolism..Other.glycan.degradation 0.11 0.01 0.20 0.0257 0.565
21 Environmental.Information.Processing..Signaling.Molecules.and.Interaction..Bacterial.toxins 0.09 0.01 0.17 0.0257 0.565
203 Environmental.Information.Processing..Membrane.Transport..Transporters 0.03 0.00 0.05 0.0293 0.565
45 Metabolism..Xenobiotics.Biodegradation.and.Metabolism..Chloroalkane.and.chloroalkene.degradation 0.04 0.00 0.08 0.0309 0.565
128 Organismal.Systems..Endocrine.System..PPAR.signaling.pathway 0.08 0.01 0.16 0.0348 0.565
80 Metabolism..Lipid.Metabolism..Glycerophospholipid.metabolism -0.03 -0.06 0.00 0.0356 0.565
81 Metabolism..Amino.Acid.Metabolism..Glycine..serine.and.threonine.metabolism 0.02 0.00 0.03 0.0362 0.565
213 Metabolism..Amino.Acid.Metabolism..Valine..leucine.and.isoleucine.biosynthesis 0.03 0.00 0.06 0.0383 0.565
2 Organismal.Systems..Endocrine.System..Adipocytokine.signaling.pathway 0.10 0.00 0.19 0.0427 0.565
50 Metabolism..Amino.Acid.Metabolism..Cysteine.and.methionine.metabolism 0.02 0.00 0.05 0.0461 0.565
182 Metabolism..Metabolism.of.Other.Amino.Acids..Selenocompound.metabolism 0.02 0.00 0.04 0.0463 0.565

The analyses for data of other studies were done similarly.

Meta-analysis of four studies

#load save results of four studies for the comparison of pathway relative abundance between genders adjusted for breastfeeding status and infant age at sample collection 
data(pathcom.unc6.rel.gamlss.sexg)
data(pathcom.ha6.rel.gamlss.sexg)
data(pathcom.rm6.rel.gamlss.sexg)
data(pathcom.usbmk6.rel.gamlss.sexg)
#Bangladesh
taxacom.zi.rm<-pathcom.rm6.rel.gamlss.sexg
for (i in 1: length(names(taxacom.zi.rm))){
  taxacom.zi.rm[[i]]<-as.data.frame(taxacom.zi.rm[[i]])
  taxacom.zi.rm[[i]][,'path']<-rownames(taxacom.zi.rm[[i]])
  taxacom.zi.rm[[i]][,'study']<-"Subramanian et al 2014 (Bangladesh)"
  taxacom.zi.rm[[i]][,'pop']<-"Bangladesh"
}
#Haiti
taxacom.zi.ha<-pathcom.ha6.rel.gamlss.sexg
for (i in 1: length(names(taxacom.zi.ha))){
  taxacom.zi.ha[[i]]<-as.data.frame(taxacom.zi.ha[[i]])
  taxacom.zi.ha[[i]][,'path']<-rownames(taxacom.zi.ha[[i]])
  taxacom.zi.ha[[i]][,'study']<-"Bender et al 2016 (Haiti)"
  taxacom.zi.ha[[i]][,'pop']<-"Haiti"
}
#CA-FL
taxacom.zi.usbmk<-pathcom.usbmk6.rel.gamlss.sexg
for (i in 1: length(names(taxacom.zi.usbmk))){
  taxacom.zi.usbmk[[i]]<-as.data.frame(taxacom.zi.usbmk[[i]])
  taxacom.zi.usbmk[[i]][,'path']<-rownames(taxacom.zi.usbmk[[i]])
  taxacom.zi.usbmk[[i]][,'study']<-"Pannaraj et al 2017 (USA(CA_FL))"
  taxacom.zi.usbmk[[i]][,'pop']<-"USA(CA_FL)"
}
#NC
taxacom.zi.unc<-pathcom.unc6.rel.gamlss.sexg
for (i in 1:length(names(taxacom.zi.unc))){ #
  taxacom.zi.unc[[i]]<-as.data.frame(taxacom.zi.unc[[i]])
  taxacom.zi.unc[[i]][,'path']<-rownames(taxacom.zi.unc[[i]])
  taxacom.zi.unc[[i]][,'study']<-"Thompson et al 2015 (USA(NC))"
  taxacom.zi.unc[[i]][,'pop']<-"USA(NC)"
}

taxacom.zi.l2<-rbind.fill(taxacom.zi.rm$l2,taxacom.zi.ha$l2,taxacom.zi.unc$l2,taxacom.zi.usbmk$l2)
#taxacom.zi.l2$pop<-as.factor(taxacom.zi.l2$pop)
taxacom.zi.l3<-rbind.fill(taxacom.zi.rm$l3,taxacom.zi.ha$l3,taxacom.zi.unc$l3,taxacom.zi.usbmk$l3)
#taxacom.zi.l3$pop<-as.factor(taxacom.zi.l3$pop)
pathcom.zi.sexg<-list(l2=taxacom.zi.l2,l3=taxacom.zi.l3)

# meta-analysis (take some time to run)
#pathmetatab.zi.sex.l2<-meta.taxa(taxcomdat=pathcom.zi.sexg$l2, sm="RR",studylab = "pop", p.adjust.method="fdr",percent.meta=0.5,pool.var="id")
#pathmetatab.zi.sex.l3<-meta.taxa(taxcomdat=pathcom.zi.sexg$l3, sm="RR",studylab = "pop", p.adjust.method="fdr",percent.meta=0.5,pool.var="id")
# load saved results 
data(pathmetatab.zi.sexg) 

#level 2
kable(metatab.show(metatab=pathmetatab.zi.sex.l2$random,com.pooled.tab=pathcom.zi.sexg$l2,sumvar="path",showvar="genderMale",p.cutoff.type="p", p.cutoff=0.05,display="table"))
id estimate ll ul p p.adjust
26 Metabolism..Metabolism.of.Terpenoids.and.Polyketides -0.01 -0.02 0 0.0417 0.9902 
# Nice plot all pathways (use different color scale for pathway)
metadat<-metatab.show(metatab=pathmetatab.zi.sex.l2$random,com.pooled.tab=pathcom.zi.sexg$l2,sumvar="path",showvar="genderMale",p.cutoff.type="p", p.cutoff=1,display="data")
metadat$taxsig.all$pop<-factor(metadat$taxsig.all$pop,levels=c("Bangladesh","Haiti","USA(CA_FL)","USA(NC)","Pooled"))
meta.niceplot(metadat=metadat,sumtype="path",p="p",p.adjust="p.adjust",p.sig.heat="yes",heat.forest.width.ratio =c(1,1.3),est.break = c(-Inf, -0.5,-0.1,-0.05,0,0.05,0.1,0.5, Inf),est.break.label = c("<-0.5)", "[-0.5,-0.1)","[-0.1,-0.05)","[-0.05,0)","[0,0.05)","[0.05,0.1)", "[0.1,0.5)",">=0.5"),leg.key.size=0.8,leg.text.size=10,heat.text.x.size=6,forest.axis.text.y=6,forest.axis.text.x=6, point.ratio = c(4,2),line.ratio = c(2,1))

#Level 3
kable(metatab.show(metatab=pathmetatab.zi.sex.l3$random,com.pooled.tab=pathcom.zi.sexg$l3,sumvar="path",showvar="genderMale",p.cutoff.type="p", p.cutoff=0.05,display="table"))
id estimate ll ul p p.adjust
40 Unclassified..Cellular.Processes.and.Signaling..Cell.division 0.08 0.01 0.16 0.0324 0.9975
98 Metabolism..Lipid.Metabolism..Lipid.biosynthesis.proteins -0.01 -0.02 0.00 0.0411 0.9975
16 Metabolism..Carbohydrate.Metabolism..Ascorbate.and.aldarate.metabolism 0.03 0.00 0.06 0.0472 0.9975 
# Nice plot for pathways with pooled p-values<=0.3
metadat<-metatab.show(metatab=pathmetatab.zi.sex.l3$random,com.pooled.tab=pathcom.zi.sexg$l3,sumvar="path",showvar="genderMale",p.cutoff.type="p", p.cutoff=0.3,display="data")
metadat$taxsig.all$pop<-factor(metadat$taxsig.all$pop,levels=c("Bangladesh","Haiti","USA(CA_FL)","USA(NC)","Pooled"))
meta.niceplot(metadat=metadat,sumtype="path",p="p",p.adjust="p.adjust",est.break = c(-Inf, -0.5,-0.1,-0.05,0,0.05,0.1,0.5, Inf),est.break.label = c("<-0.5)", "[-0.5,-0.1)","[-0.1,-0.05)","[-0.05,0)","[0,0.05)","[0.05,0.1)","[0.1,0.5)",">=0.5"),heat.forest.width.ratio=c(1,1.3),leg.key.size=1,leg.text.size=8,heat.text.x.size=6,forest.axis.text.y=6,forest.axis.text.x=6, point.ratio = c(4,2),line.ratio = c(2,1))

Meta-analysis of other microbiome measures

Random effects meta-analysis models can also be generally applied to other microbiome measures such as microbial alpha diversity and microbiome age. To make the estimates for these positive continuous microbiome measures comparable across studies, these measures should be standardized to have a mean of 0 and standard deviation of 1 before between-group-comparison within each study. Random effects meta-analysis models can then be applied to pool the “comparable” estimates and their standard errors across studies. Meta-analysis results of these measures can be displayed as standard meta-analysis forest plots.

Alpha diversity

Calculate mean alpha diversity indexes for a selected rarefaction depth, standardize and compare standardized alpha diversity indexes between groups (male vs. female infants <=6 months of age) adjusting for covariates (feeding status and infant age at sample collection) using Bangladesh data

For each study, the alpha.compare function imports the outputs from “alpha_rarefaction.py” QIIME1 script and calculates mean alpha diversity for different indices for each sample based on a user defined rarefaction depth. Mean alpha diversity indexes are standardized to have a mean of 0 and standard deviation of 1 to make these measures comparable across studies. Standardized alpha diversity indexes are compared between groups adjusting for covariates using LM. Meta-analysis across studies is then done and the results are displayed as a standard meta-analysis forest plot.

data(sam.rm)
patht<-system.file("extdata/QIIME_outputs/Bangladesh/alpha_div_collated", package = "metamicrobiomeR", mustWork = TRUE)
alpha.rm<-read.multi(patht=patht,patternt=".txt",assignt="no",study="Bangladesh")
names(alpha.rm)<-sub(patht,"",names(alpha.rm))
samfile<-merge(samde, he50[,c("child.id","gender","month.exbf","month.food")],by="child.id")
samfile$age.sample<-samfile$age.months
samfile$bf<-factor(samfile$bf,levels=c("ExclusiveBF","Non_exclusiveBF","No_BF"))
samfile$personid<-samfile$child.id
samfile$sampleid<-tolower(samfile$fecal.sample.id)
#comparison of standardized alpha diversity indexes between genders adjusting for breastfeeding and infant age at sample collection in infants <=6 months of age 
alphacom6.rm.sexsg<-alpha.compare(datlist=alpha.rm,depth=3,mapfile=samfile,mapsampleid="fecal.sample.id",comvar="gender",adjustvar=c("age.sample","bf"),longitudinal="yes",age.limit=6,standardize=TRUE)
kable(alphacom6.rm.sexsg$alphasum[,1:5])
id Estimate.genderMale Std. Error.genderMale t value.genderMale Pr(>|t|).genderMale
chao1 0.0652722 0.0721873 0.9042059 0.3658862
observed_species 0.0652941 0.0621281 1.0509593 0.2932773
pd_whole_tree 0.0313157 0.0502442 0.6232707 0.5331066
shannon -0.0012757 0.0820824 -0.0155418 0.9875999 
alpha.sexs<-merge(samfile,alphacom6.rm.sexsg$alphamean.standardized,by="sampleid")

#plot curves of standardized Shannon index by age and gender with Generalized Additive Mixed Models (GAMM) 
alpha.sexs$gender<-as.factor(alpha.sexs$gender)
alpha.sexs$bf<-as.factor(alpha.sexs$bf)
gfit<-gamm(shannon~s(age.sample,by=gender) +gender,family=gaussian,
         data=alpha.sexs,random=list(personid=~1))
pred <- predict(gfit$gam, newdata = alpha.sexs,se.fit=TRUE)
datfit<-cbind(alpha.sexs, fit=pred$fit,ul=(pred$fit+(1.96*pred$se.fit)),ll=(pred$fit-(1.96*pred$se.fit)))
ggplot()+ geom_point(data = subset(alpha.sexs,age.sample<=6), aes(x = age.sample, y = shannon, group = personid, colour=gender),size=1)+ 
  geom_line(data = subset(alpha.sexs,age.sample<=6), aes(x = age.sample, y = shannon, group = personid, colour=gender),size=0.1)+
  geom_line(data = subset(datfit,age.sample<=6),aes(x = age.sample, y = fit, colour=gender),size = 1)+
  geom_ribbon(data = subset(datfit,age.sample<=6),aes(x=age.sample, ymax=ul, ymin=ll, fill=gender), alpha=.5)+guides(fill=FALSE)+
  xlab("Chronological age (month)") +ylab("Standardized Shannon index")+
  scale_x_continuous(breaks=seq(from=0,to=24,by=3),
                     labels=seq(from=0,to=24,by=3))+
  labs(color='')+
  theme(legend.position = "right",
        axis.line = element_line(colour = "black"),
        panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        panel.border = element_blank(),
        panel.background = element_blank(),
        strip.background =element_rect(fill="white"))

The analyses for other studies were done similarly.

Meta-analysis of four studies

The results showed that alpha diversity (four commonly used indexes Shannon, Phylogenetic diversity whole tree, Observed species, Chao1) was not different between male and female infants <=6 months of age in the meta-analysis of the four included studies.

# load saved results of 4 studies 
data(alphacom6.sex4.scaledg)
# put data from 4 studies together for meta-analysis 
asum.ba<-alphacom6.rm.sexsg$alphasum
asum.ba$pop<-"Bangladesh"
asum.ha<-alphacom6.ha.sexsg$alphasum
asum.ha$pop<-"Haiti"
asum.cafl<-alphacom6.usbmk.sexsg$alphasum
asum.cafl$pop<-"USA(CA_FL)"
asum.unc<-alphacom6.unc.sexsg$alphasum
asum.unc$pop<-"USA(UNC)"
asum4<-rbind.fill(asum.ba,asum.ha,asum.cafl,asum.unc)
kable(asum4[,c(colnames(asum4)[1:5],"pop")])
id Estimate.genderMale Std. Error.genderMale t value.genderMale Pr(>|t|).genderMale pop
chao1 0.0652722 0.0721873 0.9042059 0.3658862 Bangladesh
observed_species 0.0652941 0.0621281 1.0509600 0.2932770 Bangladesh
pd_whole_tree 0.0313157 0.0502441 0.6232717 0.5331060 Bangladesh
shannon -0.0012757 0.0820824 -0.0155418 0.9875999 Bangladesh
chao1 -0.0886223 0.3528103 -0.2511895 0.8029223 Haiti
observed_species -0.0914700 0.3453005 -0.2648999 0.7924139 Haiti
pd_whole_tree 0.0425683 0.2952372 0.1441835 0.8860620 Haiti
shannon -0.0091118 0.2935656 -0.0310382 0.9753897 Haiti
chao1 -0.0450324 0.1337008 -0.3368147 0.7362566 USA(CA_FL)
observed_species -0.0653818 0.1144946 -0.5710474 0.5679675 USA(CA_FL)
pd_whole_tree -0.1590197 0.1075934 -1.4779691 0.1394160 USA(CA_FL)
shannon -0.0705346 0.1433712 -0.4919718 0.6227392 USA(CA_FL)
chao1 0.4965750 0.5924682 0.8381463 0.4019485 USA(UNC)
observed_species 0.2336717 0.5294163 0.4413761 0.6589407 USA(UNC)
pd_whole_tree -0.0107622 0.7331424 -0.0146796 0.9882878 USA(UNC)
shannon 0.1783060 0.5552779 0.3211113 0.7481260 USA(UNC) 
#Shannon index 
shannon.sex <- metagen(Estimate.genderMale, `Std. Error.genderMale`, studlab=pop,data=subset(asum4,id=="shannon"),sm="RD", backtransf=FALSE)
forest(shannon.sex,smlab="Standardized \n diversity difference",sortvar=subset(asum4,id=="shannon")$pop,lwd=2)

shannon.sex
                RD            95%-CI %W(fixed) %W(random)
Bangladesh -0.0013 [-0.1622; 0.1596]      70.0       70.0
Haiti      -0.0091 [-0.5845; 0.5663]       5.5        5.5
USA(CA_FL) -0.0705 [-0.3515; 0.2105]      23.0       23.0
USA(UNC)    0.1783 [-0.9100; 1.2666]       1.5        1.5

Number of studies combined: k = 4

                          RD            95%-CI     z p-value
Fixed effect model   -0.0149 [-0.1495; 0.1198] -0.22  0.8288
Random effects model -0.0149 [-0.1495; 0.1198] -0.22  0.8288

Quantifying heterogeneity:
tau^2 = 0; H = 1.00 [1.00; 1.00]; I^2 = 0.0% [0.0%; 0.0%]

Test of heterogeneity:
    Q d.f. p-value
 0.30    3  0.9601

Details on meta-analytical method:
- Inverse variance method
- DerSimonian-Laird estimator for tau^2
kable(cbind(study=shannon.sex$studlab,pval=shannon.sex$pval))
study pval
Bangladesh 0.987599902164408
Haiti 0.9752390484969
USA(CA_FL) 0.622739246734807
USA(UNC) 0.748126030769876 
# Other indexes
chao1.sex <- metagen(Estimate.genderMale, `Std. Error.genderMale`, studlab=pop,data=subset(asum4,id=="chao1"),sm="RD", backtransf=FALSE)
observed_species.sex <- metagen(Estimate.genderMale, `Std. Error.genderMale`, studlab=pop,data=subset(asum4,id=="observed_species"),sm="RD", backtransf=FALSE)
pd_whole_tree.sex <- metagen(Estimate.genderMale, `Std. Error.genderMale`, studlab=pop,data=subset(asum4,id=="pd_whole_tree"),sm="RD", backtransf=FALSE)
#show random meta-analysis model results of all indexes
atab<-as.data.frame(cbind(estimate=c(shannon.sex$TE.random,chao1.sex$TE.random,observed_species.sex$TE.random,pd_whole_tree.sex$TE.random),
                          ll=c(shannon.sex$lower.random,chao1.sex$lower.random,observed_species.sex$lower.random,pd_whole_tree.sex$lower.random),
                          ul=c(shannon.sex$upper.random,chao1.sex$upper.random,observed_species.sex$upper.random,pd_whole_tree.sex$upper.random),
                          index=c("shannon","chao1","observed_species","pd_whole_tree")))
atab[,1:3]<-lapply(atab[,1:3],as.character)
atab[,1:3]<-lapply(atab[,1:3],as.numeric)
a4<-ggplot(data=atab,aes(x=estimate,y=index))+
  geom_point(shape=16, colour="black")+
  geom_errorbarh(aes(xmin=ll,xmax=ul),height=0.0, colour="black")+
  geom_vline(xintercept=0,linetype="dashed")+
  scale_x_continuous(breaks=seq(from=-0.5,to=0.5,by=0.1),
                     labels=seq(from=-0.5,to=0.5,by=0.1))+
  theme(axis.line = element_line(colour = "black"),
        panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        panel.border = element_blank(),
        panel.background = element_blank())+
  xlab("Pooled standardized diversity difference")+ylab("Alpha diversity index")
a4

Microbiome age

Predicting microbiome age, checking model performance, and replicate the results of the Bangladesh study

Random Forest (RF) modeling of gut microbiota maturity has been widely used to characterize development of the microbiome over chronological time. Adapting from the original approach of Subramanian et al, in the microbiomeage function, relative abundances of bacterial genera that were detected in the Bangladesh data and in the data of other studies to be included were regressed against infant chronological age using a RF model on a predefined training dataset of the Bangladesh study. This predefined training set includes 249 samples collected monthly from birth to 2 years of age from 11 Bangladeshi healthy singleton infants. The RF training model fit based on relative abundances of these shared bacterial genera was then used to predict infant age on the test data of the Bangladesh study and the data of each other study to be included. The predicted infant age based on relative abundances of these shared bacterial genera in each study is referred to as gut microbiota age.

In brief, the microbiomeage function get the shared genera list between the Bangladesh study and all other included studies, get the training and test sets from Bangladesh data based on the shared genera list, fit the train Random Forest model and predict microbiome age in the test set of Bangladesh data and data from all included studies, check for performance of the model based on the shared genera list on Bangladesh healthy cohort data, reproduce the findings of the Bangladesh malnutrition study.

The RF model based on the relative abundance of the shared bacterial genera of the four included studies explained 96% of the variance related to chronological age in the training set and 67% of the variance related to chronological age in the test set of Bangladesh data. This performance is better than the original RF model proposed by Subramanian et al.

#load Bangladesh taxa relative abundance summary up to genus level merged with mapping file (output from QIIME)
bal6<-read.delim(system.file("extdata/QIIME_outputs/Bangladesh/tax_mapping7", "Subramanian_et_al_mapping_file_L6.txt", package = "metamicrobiomeR", mustWork = TRUE))
colnames(bal6)<-tolower(colnames(bal6))
#View(bal6)
#format for data of other studies should be similar to Bangladesh data, must have 'age.sample' variable as age of infant at stool sample collection 
# Load data of 3 other studies 
data(gtab.3stud)
names(gtab.3stud)
[1] "nc"    "ca_fl" "haiti"
#predict microbiome age on Bangladesh data and data of other three studies based on shared genera across 4 studies  
#(take time to run)
#miage<-microbiomeage(l6.relabundtab=gtab.3stud)
#load saved results 
data(miage)
# list of shared genera that are available in the Bangladesh study and other included studies 
kable(miage$sharedgenera.importance)
genera importance
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae.g__blautia 2481.8920619
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__ruminococcaceae.g__ 1906.7267503
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae.g__ 1382.4129745
k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__prevotellaceae.g__prevotella 650.0003357
k__bacteria.p__firmicutes.c__bacilli.o__bacillales.f__staphylococcaceae.g__staphylococcus 637.5007856
k__bacteria.p__firmicutes.c__clostridia.oclostridiales.f.g__ 632.9699303
k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__lactobacillaceae.g__lactobacillus 415.0586216
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__veillonellaceae.g__dialister 413.3290927
k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__pasteurellales.f__pasteurellaceae.g__haemophilus 356.5366196
k__bacteria.p__actinobacteria.c__actinobacteria.o__bifidobacteriales.f__bifidobacteriaceae.g__bifidobacterium 312.9315215
k__bacteria.p__actinobacteria.c__actinobacteria.o__actinomycetales.f__actinomycetaceae.g__actinomyces 220.2466611
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae.g__dorea 202.6234334
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae.g__.ruminococcus. 158.0250246
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae.g__coprococcus 152.5757667
k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__enterococcaceae.g__enterococcus 147.6969771
k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__enterobacteriales.f__enterobacteriaceae.g__ 134.5554227
k__bacteria.p__firmicutes.c__clostridia.oclostridiales.f.tissierellaceae..g__anaerococcus 134.5155377
k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__streptococcaceae.g__streptococcus 122.4810748
k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales.f__coriobacteriaceae.g__collinsella 117.5630377
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__veillonellaceae.g__veillonella 113.8576646
k__bacteria.p__actinobacteria.c__actinobacteria.o__actinomycetales.f__corynebacteriaceae.g__corynebacterium 107.8244500
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__clostridiaceae.g__clostridium 103.7414037
k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales.f__erysipelotrichaceae.g__ 91.3499859
k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__bacteroidaceae.g__bacteroides 91.0539556
k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__pseudomonadales.f__pseudomonadaceae.g__pseudomonas 89.6867279
k__bacteria.p__actinobacteria.c__actinobacteria.o__actinomycetales.f__micrococcaceae.g__rothia 73.2330650
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__clostridiaceae.g__ 73.1392380
k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales.f__coriobacteriaceae.g__ 70.4576219
k__bacteria.p__proteobacteria.c__betaproteobacteria.o__neisseriales.f__neisseriaceae.g__neisseria 63.0585978
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__ruminococcaceae.g__oscillospira 62.8580003
k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales.f__erysipelotrichaceae.g__.eubacterium. 61.6988728
k__bacteria.p__firmicutes.c__bacilli.o__gemellales.f__gemellaceae.g__ 46.9889764
k__bacteria.p__actinobacteria.c__coriobacteriia.o__coriobacteriales.f__coriobacteriaceae.g__atopobium 46.8921466
k__bacteria.p__fusobacteria.c__fusobacteriia.o__fusobacteriales.f__fusobacteriaceae.g__fusobacterium 46.6647271
k__bacteria.p__firmicutes.c__clostridia.oclostridiales.f.tissierellaceae..g__peptoniphilus 45.0260938
k__bacteria.p__proteobacteria.c__betaproteobacteria.o__burkholderiales.f__alcaligenaceae.g__sutterella 38.8518496
k__bacteria.p__firmicutes.c__clostridia.oclostridiales.f.tissierellaceae..g__finegoldia 33.3776433
k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__streptococcaceae.g__lactococcus 32.9668029
k__bacteria.p__firmicutes.c__bacilli.o__bacillales.f__bacillaceae.g__bacillus 30.7285110
k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__porphyromonadaceae.g__parabacteroides 29.7347955
k__bacteria.p__bacteroidetes.c__bacteroidia.o__bacteroidales.f__rikenellaceae.g__ 24.7278422
k__bacteria.p__firmicutes.c__erysipelotrichi.o__erysipelotrichales.f__erysipelotrichaceae.g__bulleidia 7.2099213
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__veillonellaceae.g__acidaminococcus 6.5118551
k__bacteria.p__cyanobacteria.c__chloroplast.ostreptophyta.f.g__ 6.1609772
k__bacteria.p__firmicutes.c__bacilli.o__lactobacillales.f__carnobacteriaceae.g__granulicatella 3.6444079
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__lachnospiraceae.g__roseburia 3.2288045
k__bacteria.p__bacteroidetes.c__flavobacteriia.oflavobacteriales.f.weeksellaceae..g__cloacibacterium 1.7980170
k__bacteria.p__proteobacteria.c__alphaproteobacteria.o__rhizobiales.f__rhizobiaceae.g__agrobacterium 1.4747628
k__bacteria.p__firmicutes.c__clostridia.o__clostridiales.f__ruminococcaceae.g__anaerotruncus 0.1673286
k__bacteria.p__firmicutes.c__bacilli.o__bacillales.f__paenibacillaceae.g__paenibacillus 0.1086412
k__bacteria.p__proteobacteria.c__alphaproteobacteria.o__sphingomonadales.f__sphingomonadaceae.g__sphingomonas 0.0342345
k__bacteria.p__proteobacteria.c__gammaproteobacteria.o__pseudomonadales.f__moraxellaceae.g__ 0.0106534 
#check performance
grid.arrange(miage$performanceplot$ptrain, miage$performanceplot$ptest,nrow=1)

#replicate the findings of Subramanian et al paper 
ggplot() +geom_point(data=miage$microbiomeage.bangladesh$all,aes(x=age.sample, y=age.predicted, colour=health_analysis_groups))

Comparison of standardized predicted microbiome age between male vs. female infants <= 6 months of age adjusting for age of infants and breastfeeding status at sample collection using Bangladesh data

The predicted infant age in each included study based on relative abundance of the shared gut bacterial genera using the above RF model is referred to as gut microbiota age.Gut microbiotat age is standardized to have a mean of 0 and standard deviation of 1. Standardized gut microbiota age are compared between groups adjusting for covariates using LM and meta-analysis across studies is then done.

samhe<-merge(samde,he50[,c("child.id","gender","month.exbf","month.food")],by="child.id")
rmdat.rm<-merge(samhe,miage$microbiomeage.bangladesh$healthy,by.y="sampleid",by.x="fecal.sample.id")
# plot curves with GAMM 
rmdat.rm$gender<-as.factor(rmdat.rm$gender)
rmdat.rm$bf<-as.factor(rmdat.rm$bf)
gfit<-gamm(age.predicted~s(age.sample,by=gender) +gender,family=gaussian,
         data=rmdat.rm,random=list(personid=~1))
pred <- predict(gfit$gam, newdata = rmdat.rm,se.fit=TRUE)
datfit<-cbind(rmdat.rm, fit=pred$fit,ul=(pred$fit+(1.96*pred$se.fit)),ll=(pred$fit-(1.96*pred$se.fit)))
ggplot()+ geom_point(data = subset(rmdat.rm,age.sample<=6), aes(x = age.sample, y = age.predicted, group = personid, colour=gender),size=1)+ 
  geom_line(data = subset(rmdat.rm,age.sample<=6), aes(x = age.sample, y = age.predicted, group = personid, colour=gender),size=0.1)+
  geom_line(data = subset(datfit,age.sample<=6),aes(x = age.sample, y = fit, colour=gender),size = 1)+
  geom_ribbon(data = subset(datfit,age.sample<=6),aes(x=age.sample, ymax=ul, ymin=ll, fill=gender), alpha=.5)+guides(fill=FALSE)+
  xlab("Chronological age (month)") +ylab("Microbiome age (month)")+
  scale_x_continuous(breaks=seq(from=0,to=24,by=3),
                     labels=seq(from=0,to=24,by=3))+
  labs(color='')+
  theme(legend.position = "right",
        axis.line = element_line(colour = "black"),
        panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        panel.border = element_blank(),
        panel.background = element_blank(),
        strip.background =element_rect(fill="white"))

rmdat.rm$personid<-paste("rm",as.factor(tolower(rmdat.rm$personid)),sep=".")
rmdat.rm$sampleid<-paste("rm",tolower(rmdat.rm$fecal.sample.id),sep=".")
rmdat.rm$author<-"Subramanian et al"
rmdat.rm$pop<-"Bangladesh"
rmdat.rm$year<-"2014"
# standardize age.predicted to have mean of zero and standard deviation of 1 
rmdat.rm$age.predicteds<-(rmdat.rm$age.predicted-mean(rmdat.rm$age.predicted,na.rm=T))/sd(rmdat.rm$age.predicted)
# Comparison in infants <=6 months of age 
fitsum<-summary(lmer(age.predicteds~gender+bf+age.sample+(1|personid),data=subset(rmdat.rm,age.sample<=6)))
fitdat<-as.data.frame(fitsum$coefficients[-1,])
fitdat[,"varname"]<-rownames(fitdat)
fitdat[,"pop"]<-"Bangladesh"
kable(fitdat)
Estimate Std. Error df t value Pr(>|t|) varname pop
genderMale -0.0909496 0.0595166 36.06456 -1.5281378 0.1352025 genderMale Bangladesh
bfNo_BF -0.1075706 0.1666272 248.75546 -0.6455766 0.5191486 bfNo_BF Bangladesh
bfNon_exclusiveBF 0.0339266 0.0571267 202.03700 0.5938845 0.5532537 bfNon_exclusiveBF Bangladesh
age.sample 0.0862128 0.0136845 236.02555 6.3000537 0.0000000 age.sample Bangladesh 
rm.ba.sex<-reshape(fitdat, idvar="pop", timevar="varname", direction="wide")
kable(rm.ba.sex)
pop Estimate.genderMale Std. Error.genderMale df.genderMale t value.genderMale Pr(>|t|).genderMale Estimate.bfNo_BF Std. Error.bfNo_BF df.bfNo_BF t value.bfNo_BF Pr(>|t|).bfNo_BF Estimate.bfNon_exclusiveBF Std. Error.bfNon_exclusiveBF df.bfNon_exclusiveBF t value.bfNon_exclusiveBF Pr(>|t|).bfNon_exclusiveBF Estimate.age.sample Std. Error.age.sample df.age.sample t value.age.sample Pr(>|t|).age.sample
genderMale Bangladesh -0.0909496 0.0595166 36.06456 -1.528138 0.1352025 -0.1075706 0.1666272 248.7555 -0.6455766 0.5191486 0.0339266 0.0571267 202.037 0.5938845 0.5532537 0.0862128 0.0136845 236.0255 6.300054 0

The analyses for other studies were done similarly.

Meta-analysis of four studies

The results showed that standardized microbiota age was significantly different between males vs. females but in opposite directions in two studies with small sample sizes (Haiti and North Carolina). However, meta-analysis of all four studies revealed no significant difference in gut microbiota age between genders after adjusting for feeding status and infant age at time of sample collection.

#load saved results of four studies
data(rm4.sexs)
kable(rm4.sexs)
pop Estimate.genderMale Std. Error.genderMale df.genderMale t value.genderMale Pr(>|t|).genderMale Estimate.bfNon_exclusiveBF Std. Error.bfNon_exclusiveBF df.bfNon_exclusiveBF t value.bfNon_exclusiveBF Pr(>|t|).bfNon_exclusiveBF Estimate.bfNo_BF Std. Error.bfNo_BF df.bfNo_BF t value.bfNo_BF Pr(>|t|).bfNo_BF Estimate.age.sample Std. Error.age.sample df.age.sample t value.age.sample Pr(>|t|).age.sample
Bangladesh -0.0982571 0.0635555 36.765259 -1.5460054 0.1306687 0.0466096 0.0598532 207.991061 0.7787320 0.4370226 -0.0752122 0.1738326 248.95022 -0.432670 0.6656291 0.0895102 0.0142403 235.73826 6.285695 0.0000000
Haiti 0.5595183 0.2722964 NA 2.0548132 0.0461522 -0.3112344 0.3232793 NA -0.9627416 0.3411877 NA NA NA NA NA 0.2466495 0.0802007 NA 3.075402 0.0036890
USA(CA_FL) -0.1007744 0.1133121 69.802832 -0.8893527 0.3768683 0.2476086 0.1134385 148.116702 2.1827570 0.0306289 0.9031475 0.2439538 171.39469 3.702126 0.0002880 0.2696717 0.0282094 205.87222 9.559643 0.0000000
USA(NC) -0.7286074 0.3616621 3.088779 -2.0146081 0.1347374 1.0076149 0.3132025 5.097796 3.2171356 0.0229018 -0.8707151 0.6762228 11.38869 -1.287616 0.2234301 0.4687383 0.0845192 15.95532 5.545941 0.0000447 
rm.sex<-metagen(Estimate.genderMale, `Std. Error.genderMale`, studlab=pop,data=rm4.sexs,sm="RD", backtransf=FALSE)
forest(rm.sex,smlab="Standardized \n microbiome age difference",lwd=2)

rm.sex
                RD             95%-CI %W(fixed) %W(random)
Bangladesh -0.0983 [-0.2228;  0.0263]      71.4       40.7
Haiti       0.5595 [ 0.0258;  1.0932]       3.9       15.4
USA(CA_FL) -0.1008 [-0.3229;  0.1213]      22.5       33.7
USA(NC)    -0.7286 [-1.4375; -0.0198]       2.2       10.2

Number of studies combined: k = 4

                          RD            95%-CI     z p-value
Fixed effect model   -0.0871 [-0.1924; 0.0181] -1.62  0.1048
Random effects model -0.0625 [-0.3203; 0.1953] -0.48  0.6348

Quantifying heterogeneity:
tau^2 = 0.0385; H = 1.72 [1.00; 2.94]; I^2 = 66.0% [0.4%; 88.4%]

Test of heterogeneity:
    Q d.f. p-value
 8.83    3  0.0316

Details on meta-analytical method:
- Inverse variance method
- DerSimonian-Laird estimator for tau^2
kable(cbind(study=rm.sex$studlab,pval=rm.sex$pval))
study pval
Bangladesh 0.122103269327307
Haiti 0.0398970499413607
USA(CA_FL) 0.373813543114747
USA(NC) 0.0439457250978759

Discussion and Conclusion

Our metamicrobiomeR package implemented GAMLSS-BEZI for analysis of microbiome relative abundance data and random effect meta-analysis models for meta-analysis across microbiome studies. The advantages of GAMLSS-BEZI are: 1) it directly and properly address the distribution of microbiome relative abundance data which resemble a zero-inflated beta distribution; 2) it has better power to detect differential relative abundances between groups than the commonly used approach LMAS; 3) the estimates from GAMLSS-BEZI are log(odds ratio) of relative abundances between groups and thus are comparable across studies. Random effects meta-analysis models can be directly applied to pool these adjusted estimates and their standard errors across studies. This approach allows examination of study-specific effects, heterogeneity between studies, and the overall pooled effects across microbiome studies. Besides, random effects meta-analysis models can also generally applied to other microbiome measures such as diversity indexes or microbiome age. Standardization of these measures before comparison between groups within each study also make the estimates for these measures comparable across studies. The examples and workflow using our metamicrobiomeR package are reproducible and applicable for the analysis and meta-analysis of other microbiome studies.

Availability

All source code, example data, documentation and the manuscript describing the metamicrobiomeR package are available at [https://github.com/nhanhocu/metamicrobiomeR].

Funding

This work was supported by Mervyn W. Susser fellowship in the Gertrude H. Sergievsky Center, Columbia University Medical Center (to Nhan Thi Ho) during the development and supported by Vinmec Healthcare System, Vietnam (to Nhan Thi Ho) during the revision.

Section information

sessionInfo()
R version 3.5.2 (2018-12-20)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 17134)

Matrix products: default

locale:
[1] LC_COLLATE=English_United States.1252 
[2] LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] meta_4.9-4          mgcv_1.8-27         nlme_3.1-137       
 [4] lmerTest_3.1-0      lme4_1.1-20         Matrix_1.2-15      
 [7] ggplot2_3.1.0       gridExtra_2.3       gdata_2.18.0       
[10] dplyr_0.8.0.1       plyr_1.8.4          metamicrobiomeR_1.1
[13] usethis_1.4.0       devtools_2.0.1      knitr_1.21         

loaded via a namespace (and not attached):
 [1] pkgload_1.0.2      splines_3.5.2      foreach_1.4.4     
 [4] prodlim_2018.04.18 gtools_3.8.1       assertthat_0.2.0  
 [7] stats4_3.5.2       highr_0.7          yaml_2.2.0        
[10] remotes_2.0.2      ipred_0.9-8        sessioninfo_1.1.1 
[13] numDeriv_2016.8-1  pillar_1.3.1       backports_1.1.3   
[16] lattice_0.20-38    glue_1.3.0         digest_0.6.18     
[19] RColorBrewer_1.1-2 minqa_1.2.4        colorspace_1.4-0  
[22] recipes_0.1.4      htmltools_0.3.6    timeDate_3043.102 
[25] pkgconfig_2.0.2    caret_6.0-81       purrr_0.3.0       
[28] scales_1.0.0       processx_3.2.1     gower_0.1.2       
[31] lava_1.6.5         tibble_2.0.1       generics_0.0.2    
[34] withr_2.1.2        nnet_7.3-12        lazyeval_0.2.1    
[37] cli_1.0.1          survival_2.43-3    magrittr_1.5      
[40] crayon_1.3.4       memoise_1.1.0      evaluate_0.13     
[43] ps_1.3.0           fs_1.2.6           MASS_7.3-51.1     
[46] class_7.3-14       pkgbuild_1.0.2     tools_3.5.2       
[49] data.table_1.12.0  prettyunits_1.0.2  stringr_1.4.0     
[52] munsell_0.5.0      callr_3.1.1        compiler_3.5.2    
[55] rlang_0.3.1        grid_3.5.2         nloptr_1.2.1      
[58] iterators_1.0.10   labeling_0.3       rmarkdown_1.11    
[61] gtable_0.2.0       ModelMetrics_1.2.2 codetools_0.2-15  
[64] curl_3.3           reshape2_1.4.3     R6_2.4.0          
[67] lubridate_1.7.4    rprojroot_1.3-2    desc_1.2.0        
[70] stringi_1.3.1      Rcpp_1.0.0         rpart_4.1-13      
[73] tidyselect_0.2.5   xfun_0.5

References

  1. Rigby RA, Stasinopoulos DM. Generalized additive models for location, scale and shape (with discussion). J R Stat Soc Ser C (Applied Stat). 2005;54:507-54.↩

  2. Subramanian S, Huq S, Yatsunenko T, Haque R, Mahfuz M, Alam MA, et al. Persistent gut microbiota immaturity in malnourished Bangladeshi children. Nature. 2014;510:417-21.↩

  3. Bender JM, Li F, Martelly S, Byrt E, Rouzier V, Leo M, et al. Maternal HIV infection influences the microbiome of HIV-uninfected infants. Sci Transl Med. 2016;8:349ra100.↩

  4. Pannaraj PS, Li F, Cerini C, Bender JM, Yang S, Rollie A, et al. Association Between Breast Milk Bacterial Communities and Establishment and Development of the Infant Gut Microbiome. JAMA Pediatr. 2017;90095:647-54.↩

  5. Thompson AL, Monteagudo-Mera A, Cadenas MB, Lampl ML, Azcarate-Peril MA. Milk- and solid-feeding practices and daycare attendance are associated with differences in bacterial diversity, predominant communities, and metabolic and immune function of the infant gut microbiome. Front Cell Infect Microbiol. 2015;5:3.↩