# De Kovel et al, 2018
# Scripts acommpanying Data Descriptor
# Transcriptomic analysis of left-right differences in human embryonic forebrain and midbrain


#fastq quality control FastQC v0.11.5

fastq ${file}

#Aligning with Hisat2 v2.2.0.4

indexdir="/dir/to/resources/Homo_sapiens/UCSC/hg38/hg38/"
indexbase=genome
fqdir="/dir/to/fastq/files/"
bamdir="/dir/to/Bams/"
tissue=
sample=

hisat2 -x ${indexdir}${indexbase} -1 ${fqdir}S${sample}_1.fq.gz -2 ${fqdir}S${sample}_2.fq.gz -S ${bamdir}${tissue}/${sample}.sam -p 4

#Gene counting with RSEM (version 1.3.0), using bowtie2 v2.2.6
tissue=
sample=
Fastqdir=/dir/to/fastq/files/
Outdir=/output/dir/
Outfile=${sample}
Ref=/dir/to/resources/Homo_sapiens/UCSC/hg38/hg38/GCF_GRCh38.refseq_rsem

./rsem-calculate-expression  --paired-end --bowtie2 -p 4  --append-names  ${Fastqdir}${sample}_1.fastq.gz ${Fastqdir}${sample}_2.fastq.gz ${Ref} ${Outdir}${sample}_ucsc_genecount


#--------------------------------------------------------------------------------
## Processing data with R-package edgeR, limma
# R
library(edgeR)
library(limma)

# Read in expression counts to include in dge object
genes<-read.table("Forebrain_genes_ex.txt", header=T)
counts<-read.delim("Forebrain_expression_counts_ex.txt", header=T, na.strings="NA")
rownames(counts)<-counts$gene_id
counts$gene_id<-NULL

# replace missing values with 0
counts[is.na(counts)]<-0

# Create dge object
dge <- DGEList(counts=counts, genes=genes,group=side)
dge <- calcNormFactors(dge)

# filter out lowly expressed genes using the following commands:
keep <- rowSums(cpm(dge)>5) >= 3
dge <- dge[keep, , keep.lib.sizes=FALSE]

dge$samples$lib.size<-colSums(dge$counts)

#after filter calculate normalisation again
dge <- calcNormFactors(dge)

# compute expression as log2(cpm) and write to file
log2cpm<-log2(normalized.counts)
write.table(log2cpm, "BGI_2Tissues_expression_log2.txt", row.names=T, col.names=T, sep="\t")