####Load packages needed for analysis####
library(readr)
library(maftools)
library(haven)
####Load datasets #####
LMMC_Oligo_V <- read_delim("Todas_LMMCsOligos_revisado.tsv", 
                                "\t", escape_double = FALSE, trim_ws = TRUE)

LMMC_Oligo_CL <- read_sav("LMMC y Oligo 31.01.20 (40 oligos, 23 SMD, 14 NPMs, 56 LMMCs y 102 reactivas).sav")
colnames(LMMC_Oligo_CL)[5] <- "Tumor_Sample_Barcode"
####Recode variables to fit in maftools object####
LMMC_Oligo_V$Sample <- gsub("QIAseq-DNA-smCounter2.", "", LMMC_Oligo_V$Sample)
LMMC_Oligo_V$Sample <- gsub(".smCounter.anno", "", LMMC_Oligo_V$Sample)
LMMC_Oligo_V$Sample <- gsub("^.*\\.", "", LMMC_Oligo_V$Sample)
LMMC_Oligo_V$Sample <- gsub("_.*$", "", LMMC_Oligo_V$Sample)

colnames(LMMC_Oligo_V)[2] <- "Hugo_Symbol"
colnames(LMMC_Oligo_V)[3] <- "Chromosome"
colnames(LMMC_Oligo_V)[4] <- "Start_Position"
LMMC_Oligo_V$End_Position <- LMMC_Oligo_V$Start_Position
LMMC_Oligo_V$Reference_Allele <- gsub(">.*$", "", LMMC_Oligo_V$Variant)
LMMC_Oligo_V$Tumor_Seq_Allele2 <- gsub("^.*>.*\\/", "", LMMC_Oligo_V$Variant)
colnames(LMMC_Oligo_V)[19] <- "Variant_Classification"
colnames(LMMC_Oligo_V)[9] <- "Variant_Type"
colnames(LMMC_Oligo_V)[1] <- "Tumor_Sample_Barcode"
unique(LMMC_Oligo_V$Variant_Classification)
LMMC_Oligo_V$Variant_Classification[grepl("missense_variant", LMMC_Oligo_V$Variant_Classification)] <- "Missense_Mutation"
LMMC_Oligo_V$Variant_Classification[grepl("stop_gained", LMMC_Oligo_V$Variant_Classification)] <- "Nonsense_Mutation"
LMMC_Oligo_V$Variant_Classification[grepl("frameshift_variant", LMMC_Oligo_V$Variant_Classification)] <- "Frame_Shift_Ins"
LMMC_Oligo_V$Variant_Classification[grepl("splice_acceptor_variant", LMMC_Oligo_V$Variant_Classification)] <- "Splice_Site"
LMMC_Oligo_V$Variant_Classification[grepl("splice_donor_variant", LMMC_Oligo_V$Variant_Classification)] <- "Splice_Site"
LMMC_Oligo_V$Variant_Classification[grepl("inframe_deletion", LMMC_Oligo_V$Variant_Classification)] <- "In_Frame_Del"

LMMC_Oligo_V$Variant_Type[grepl("snv", LMMC_Oligo_V$Variant_Type)] <- "SNP"
LMMC_Oligo_V$Variant_Type[grepl("deletion", LMMC_Oligo_V$Variant_Type)] <- "DEL"
LMMC_Oligo_V$Variant_Type[grepl("insertion", LMMC_Oligo_V$Variant_Type)] <- "INS"
LMMC_Oligo_V$Variant_Type[grepl("mnp", LMMC_Oligo_V$Variant_Type)] <- "SNP"


unique(LMMC_Oligo_V$Classification)
LMMC_Oligo_2 <- LMMC_Oligo_V[LMMC_Oligo_V$Classification == "Pathogenic" |
                               LMMC_Oligo_V$Classification == "Presumed Pathogenic",]

# genes <- c("ASXL1", "CALR", "CEBPA", "CSF3R", "DNMT3A", "ETV6", "EZH2", "IDH1",
#            "IDH2", "JAK2", "KIT", "KRAS", "MPL", "NRAS", "PRPF8", "RUNX1",
#            "SETBP1", "SF3B1", "SH2B3", "SRSF2", "STAG2", "TET2", "TP53", "U2AF1",
#            "ZRSR2")
###Specify shaed genes per panel ####
genes <- c('ASXL1', 'CALR', 'CBL', 'CSF3R', 'DNMT3A', 'ETV6', 'EZH2', 'IDH1', 
  'IDH2', 'JAK2', 'KIT', 'KRAS', 'MPL', 'NRAS', 'PRPF8', 'RUNX1', 'SETBP1',
  'SF3B1', 'SH2B3', 'SRSF2', 'STAG2', 'TET2', 'TP53', 'U2AF1', 'ZRSR2', 'CBL_FALSE')
genes2 <- c('ASXL1', 'CALR', 'CBL', 'CSF3R', 'DNMT3A', 'ETV6', 'EZH2', 'IDH1', 
           'IDH2', 'JAK2', 'KIT', 'KRAS', 'MPL', 'NRAS', 'PRPF8', 'RUNX1', 'SETBP1',
           'SF3B1', 'SH2B3', 'SRSF2', 'STAG2', 'TET2', 'TP53', 'U2AF1', 'ZRSR2')
library(dplyr)
LMMC_Oligo_2 <- LMMC_Oligo_2[LMMC_Oligo_2$Hugo_Symbol %in% genes,]
LMMC_Oligo_CL$clasificacion <- factor(LMMC_Oligo_CL$clasificacion, 
                                      labels = c("Oligo", "LMMC", "SMD", "NPMS", 
                                                 "Reactivos", "Otras"))

colnames(LMMC_Oligo_CL)[11] <- "Study_group"

library(plyr); library(dplyr)
multihit_data <- ddply(LMMC_Oligo_2,.(Tumor_Sample_Barcode,Hugo_Symbol),nrow)
colnames(multihit_data)[3] <- "MultiHit"

LMMC_Oligo_2 <- left_join(LMMC_Oligo_2, multihit_data)
#LMMC_Oligo_2$MultiHit[LMMC_Oligo_2$MultiHit > 1] <- "2"

LMMC_Oligo_CL$CG_alt <- factor(LMMC_Oligo_CL$CG_alt, levels = c(0,1), labels = c("No", "Yes"))
LMMC_Oligo_CL$IFSPmonosMO1 <- factor(LMMC_Oligo_CL$IFSPmonosMO1, levels = c(1,2), 
                                     labels = c("MO1>94%", "MO1<94%"))

####Generate maftools objects and plotting ####
LMMC_Oligo_MAF <- read.maf(LMMC_Oligo_2, clinicalData = LMMC_Oligo_CL)

library(RColorBrewer)
vc_cols = RColorBrewer::brewer.pal(n = 8, name = 'Paired')
vc_cols = c("#A6CEE3","darkolivegreen3", "cornflowerblue", "#33A02C","coral","#E31A1C","#FDBF6F", "#FF7F00")
names(vc_cols) = c(
  'Frame_Shift_Del',
  'Missense_Mutation',
  'Nonsense_Mutation',
  'Multi_Hit',
  'Frame_Shift_Ins',
  'In_Frame_Ins',
  'Splice_Site',
  'In_Frame_Del'
)
CL_colors = c("#D53E4F", "#99D594")
CL_colors = RColorBrewer::brewer.pal(n = 6,name = 'Spectral')
names(CL_colors) = c("Oligo", "LMMC")
CL_colors = list("Study group" = CL_colors)

pdf("Summary_all.pdf", width = 14)
plotmafSummary(maf = LMMC_Oligo_MAF, color = vc_cols, rmOutlier = TRUE, 
               addStat = 'median', dashboard = TRUE, titvRaw = FALSE)
dev.off()
pdf("Oncoplot_all_v2.pdf", width = 14)
oncoplot(maf = LMMC_Oligo_MAF, top = 25, colors = vc_cols, 
#         annotationColor = CL_colors, 
         clinicalFeatures = c("Study_group", "IFSPmonosMO1", "CG_alt"),
         additionalFeature = c("MultiHit", "2"),
         sortByAnnotation = TRUE, genes = genes2,removeNonMutated = FALSE)
dev.off()

pdf("Oncoplot_all_with_names.pdf", width = 14)
oncoplot(maf = LMMC_Oligo_MAF, top = 25, colors = vc_cols, 
         #annotationColor = CL_colors, 
         sortByAnnotation = TRUE, genes = genes2,
         removeNonMutated = FALSE, 
         showTumorSampleBarcodes = TRUE, SampleNamefontSize = 0.7,
         clinicalFeatures = c("Study_group", "IFSPmonosMO1", "CG_alt"),
         additionalFeature = list(c("MultiHit", "2"), c("MultiHit", "3"),
                                  c("MultiHit", "4"), c("MultiHit", "5")),
         additionalFeatureCol = c("white", "gray80", "gray60", "gray40"))
dev.off()

pdf("SomaticInteractions_all.pdf")
somaticInteractions(maf = LMMC_Oligo_MAF, pvalue = c(0.05, 0.1), returnAll = TRUE,
                    geneOrder = NULL, fontSize = 0.8, showSigSymbols = TRUE,genes = genes,
                    showCounts = FALSE, countStats = 'sig', countType = 'all',
                    countsFontSize = 10, countsFontColor = "white", colPal = "BrBG")
dev.off()


LMMC_samples <- LMMC_Oligo_CL$Tumor_Sample_Barcode[LMMC_Oligo_CL$Study_group == "LMMC"]
Oligo_samples <- LMMC_Oligo_CL$Tumor_Sample_Barcode[LMMC_Oligo_CL$Study_group == "Oligo"]
all_samples <- c(LMMC_samples, Oligo_samples)
setdiff(all_samples, LMMC_Oligo_2$Tumor_Sample_Barcode)

LMMC_V <- LMMC_Oligo_2[LMMC_Oligo_2$Tumor_Sample_Barcode %in% LMMC_samples,]
Oligo_V <- LMMC_Oligo_2[LMMC_Oligo_2$Tumor_Sample_Barcode %in% Oligo_samples,]

LMMC_MAF <- read.maf(LMMC_V, clinicalData = LMMC_Oligo_CL)
unique(LMMC_V$Tumor_Sample_Barcode)
Oligo_MAF <- read.maf(Oligo_V, clinicalData = LMMC_Oligo_CL)

CL_colors = c("#D53E4F", "#99D594")
CL_colors = RColorBrewer::brewer.pal(n = 4,name = 'Spectral')
names(CL_colors) = c("MO1<94%", "MO1>94%", "Yes", "No")
CL_colors = list("IFSPmonosMO1" = CL_colors[1:2], "CG_alt" = CL_colors[3:4] )

pdf("Oncoplot_LMMC.pdf", width = 14)
oncoplot(maf = LMMC_MAF, top = 25, colors = vc_cols, 
         annotationColor = CL_colors,
         sortByAnnotation = TRUE, genes = genes2,
         clinicalFeatures = c("IFSPmonosMO1", "CG_alt"),
         removeNonMutated = FALSE, 
         showTitle = FALSE,
         additionalFeature = list(c("MultiHit", "2"), c("MultiHit", "3"),
                                  c("MultiHit", "4"), c("MultiHit", "5")),
         additionalFeatureCol = c("white", "gray80", "gray60", "gray40"),
         additionalFeaturePch = c(15,16,17,18))
dev.off()

pdf("Oncoplot_Oligo.pdf", width = 14)
oncoplot(maf = Oligo_MAF, top = 25, colors = vc_cols, 
         annotationColor = CL_colors, 
         sortByAnnotation = TRUE, genes = genes2,
         clinicalFeatures = c("IFSPmonosMO1", "CG_alt"),
         removeNonMutated = FALSE,
         showTitle = FALSE,
         additionalFeature = list(c("MultiHit", "2"), c("MultiHit", "3"),
                                  c("MultiHit", "4"), c("MultiHit", "5")),
         additionalFeatureCol = c("white", "gray80", "gray60", "gray40"),
         additionalFeaturePch = c(15,16,17,18))
dev.off()

coOncoplot(m1 = LMMC_MAF, m2 = Oligo_MAF, m1Name = 'LMMC', m2Name = 'Oligo', colors = vc_cols, removeNonMutated = TRUE)

LMMC_vs_Oligo <- mafCompare(m1 = LMMC_MAF, m2 = Oligo_MAF, m1Name = 'LMMC', m2Name = 'Oligo', minMut = 5)
dev.off()
forestPlot(mafCompareRes = LMMC_vs_Oligo, pVal = 0.2, color = c('royalblue', 'maroon'), geneFontSize = 0.8)
dev.off()
pdf("Lollipop.pdf")
lollipopPlot2(m1 = LMMC_MAF, m2 = Oligo_MAF, gene = "TET2", AACol1 = "HGVSp", AACol2 = "HGVSp", m1_name = "LMMC", m2_name = "Oligo")
lollipopPlot2(m1 = LMMC_MAF, m2 = Oligo_MAF, gene = "SRSF2", AACol1 = "HGVSp", AACol2 = "HGVSp", m1_name = "LMMC", m2_name = "Oligo")
lollipopPlot2(m1 = LMMC_MAF, m2 = Oligo_MAF, gene = "ASXL1", AACol1 = "HGVSp", AACol2 = "HGVSp", m1_name = "LMMC", m2_name = "Oligo")
lollipopPlot2(m1 = LMMC_MAF, m2 = Oligo_MAF, gene = "SF3B1", AACol1 = "HGVSp", AACol2 = "HGVSp", m1_name = "LMMC", m2_name = "Oligo")
lollipopPlot2(m1 = LMMC_MAF, m2 = Oligo_MAF, gene = "ZRSR2", AACol1 = "HGVSp", AACol2 = "HGVSp", m1_name = "LMMC", m2_name = "Oligo")
lollipopPlot2(m1 = LMMC_MAF, m2 = Oligo_MAF, gene = "DNMT3A", AACol1 = "HGVSp", AACol2 = "HGVSp", m1_name = "LMMC", m2_name = "Oligo")
lollipopPlot2(m1 = LMMC_MAF, m2 = Oligo_MAF, gene = "KRAS", AACol1 = "HGVSp", AACol2 = "HGVSp", m1_name = "LMMC", m2_name = "Oligo")
lollipopPlot2(m1 = LMMC_MAF, m2 = Oligo_MAF, gene = "NRAS", AACol1 = "HGVSp", AACol2 = "HGVSp", m1_name = "LMMC", m2_name = "Oligo")
dev.off()

pdf("SomaticInteractions_LMMC.pdf")
somaticInteractions(maf = LMMC_MAF, pvalue = c(0.05, 0.1), returnAll = TRUE,
                    geneOrder = NULL, fontSize = 0.8, showSigSymbols = TRUE,genes = genes,
                    showCounts = FALSE, countStats = 'sig', countType = 'all',
                    countsFontSize = 10, countsFontColor = "white", colPal = "BrBG")
dev.off()

pdf("SomaticInteractions_Oligo.pdf")
somaticInteractions(maf = Oligo_MAF, pvalue = c(0.05, 0.1), returnAll = TRUE,
                    geneOrder = NULL, fontSize = 0.8, showSigSymbols = TRUE,genes = genes,
                    showCounts = FALSE, countStats = 'sig', countType = 'all',
                    countsFontSize = 10, countsFontColor = "white", colPal = "BrBG")
dev.off()

fab.ce = clinicalEnrichment(maf = LMMC_Oligo_MAF, clinicalFeature = 'Study_group')
fab.ce$groupwise_comparision[p_value < 0.05]

plotEnrichmentResults(enrich_res = fab.ce, pVal = 0.05)

#Split TET2 mutations
LMMC_Oligo_TET <- LMMC_Oligo_2
LMMC_Oligo_TET$Hugo_Symbol[LMMC_Oligo_TET$Hugo_Symbol == "TET2" & 
                             LMMC_Oligo_TET$Variant_Classification == "Missense_Mutation"] <- "TET2_b"
LMMC_Oligo_TET$Hugo_Symbol[LMMC_Oligo_TET$Hugo_Symbol == "TET2" & 
                             LMMC_Oligo_TET$Variant_Classification == "Nonsense_Mutation"] <- "TET2_a"
LMMC_Oligo_TET$Hugo_Symbol[LMMC_Oligo_TET$Hugo_Symbol == "TET2" & 
                             LMMC_Oligo_TET$Variant_Classification == "Frame_Shift_Ins"] <- "TET2_c"
LMMC_Oligo_TET$Hugo_Symbol[LMMC_Oligo_TET$Hugo_Symbol == "TET2" & 
                             LMMC_Oligo_TET$Variant_Classification == "Splice_Site"] <- "TET2_s"
LMMC_Oligo_TET$Hugo_Symbol[LMMC_Oligo_TET$Hugo_Symbol == "TET2" & 
                             LMMC_Oligo_TET$Variant_Classification == "In_Frame_Del"] <- "TET2_id"

LMMC_V_TET <- LMMC_Oligo_TET[LMMC_Oligo_TET$Tumor_Sample_Barcode %in% LMMC_samples,]
Oligo_V_TET <- LMMC_Oligo_TET[LMMC_Oligo_TET$Tumor_Sample_Barcode %in% Oligo_samples,]

LMMC_TET <- read.maf(LMMC_V_TET, clinicalData = LMMC_Oligo_CL)
Oligo_TET <- read.maf(Oligo_V_TET, clinicalData = LMMC_Oligo_CL)

pdf("Oncostrip_TET2_LMMC.pdf", width = 35)
oncostrip(maf = LMMC_TET, genes = c('TET2_a','TET2_b', 'TET2_c', 'TET2_s'), colors = vc_cols,
          removeNonMutated = FALSE, fontSize = 1.5, showTitle = FALSE)
dev.off()
pdf("Oncostrip_TET2_Oligo.pdf", width = 35)
oncostrip(maf = Oligo_TET, genes = c('TET2_a','TET2_b', 'TET2_c', 'TET2_s'), colors = vc_cols,
          removeNonMutated = FALSE, fontSize = 1.5, showTitle = FALSE)
dev.off()