This report has goseq results for NGT versus T2D when:

1. Top 1000 genes are marked as differentially expressed
2. Top genes with positive effect in top 1000 overall genes are marked as differentially expressed
3. Top genes with negative effect in top 1000 overall genes are marked as differentially expressed

This report was generated on June 21 2015

Goseq results also saved in csv files located on snowwhite in directory: /net/snowwhite/home/beckandy/tissue/datafreeze4/goseq/jun3/12junReps/csv

Step 1: Load in all the necessary data/libraries

library(goseq)

## Loading required package: BiasedUrn
## Loading required package: geneLenDataBase
## Loading required package: DBI

library(qvalue)

fName <- "/net/snowwhite/home/beckandy/tissue/datafreeze4/goseq/ngt_vs_t2d/peer_k03_all_genes.txt" 
outFile <- "ngt_t2d"

data <- read.table(fName, as.is=T, header=T)

gene_length_file <- "/net/snowwhite/home/beckandy/tissue/datafreeze4/goseq/jun3/length.composite.gene.models.gencode.v19"
gene_lengths = read.table(gene_length_file, header=T, as.is=T);
gene_lengths$gene = sapply(gene_lengths$gene, function(x){ unlist(strsplit(x, split="[.]"))[1] });

data$gene <- sapply(data$gene, function(x){ unlist(strsplit(x, split="[.]"))[1] });

data <- merge(data, gene_lengths, by="gene", all.x=T)
data <- data[order(data$p.value),]
data$q.value <- qvalue(data$p.value)$qvalues
data$rank <- seq(1,length(data[,1]))

minRow <- 20

Step 2: Create genes vectors

The first vector simply marks the top 1000 genes as differentially expressed. The second and third vectors mark the genes with positive or negative effect in the top 1000 as differentially expressed.

genes <- as.numeric(data$rank <= 1000)
genesPos <- as.numeric(data$rank <= 1000 & data$effect > 0)
genesNeg <- as.numeric(data$rank <= 1000 & data$effect < 0)

names(genes) <- data$gene
names(genesPos) <- data$gene
names(genesNeg) <- data$gene

There are 511 DE genes with postive effect and 489 DE genes with negative effect.

Step 3: PWFs

pwf <- nullp(genes,"hg19","ensGene",bias.data=data$length, plot.fit=FALSE)

pwfPos=nullp(genesPos,"hg19","ensGene",bias.data=data$length, plot.fit=FALSE)

pwfNeg=nullp(genesNeg,"hg19","ensGene",bias.data=data$length, plot.fit=FALSE)

Step 4: run goseq

go <- goseq(pwf,"hg19","ensGene",test.cats=c("GO:BP"))
goPos <- goseq(pwfPos,"hg19","ensGene",test.cats=c("GO:BP"))
goNeg <- goseq(pwfNeg,"hg19","ensGene",test.cats=c("GO:BP"))

rownames(go) <- NULL
rownames(goPos) <- NULL
rownames(goNeg) <- NULL

# Fix problem with some p-values being slightly more than 1
go$over_represented_pvalue[go$over_represented_pvalue>1]=1;
go$under_represented_pvalue[go$under_represented_pvalue>1]=1;
goPos$over_represented_pvalue[goPos$over_represented_pvalue>1]=1;
goPos$under_represented_pvalue[goPos$under_represented_pvalue>1]=1;
goNeg$over_represented_pvalue[goNeg$over_represented_pvalue>1]=1;
goNeg$under_represented_pvalue[goNeg$under_represented_pvalue>1]=1;

go$q.value <- qvalue(go$over_represented_pvalue)$qvalues
goPos$q.value=qvalue(goPos$over_represented_pvalue)$qvalues
goNeg$q.value=qvalue(goNeg$over_represented_pvalue)$qvalues

go$q.value2 <- qvalue(go$under_represented_pvalue)$qvalues
goPos$q.value2=qvalue(goPos$under_represented_pvalue)$qvalues
goNeg$q.value2=qvalue(goNeg$under_represented_pvalue)$qvalues

go <- go[which(go$numInCat < 1000),]
goPos <- goPos[which(goPos$numInCat < 1000),]
goNeg <- goNeg[which(goNeg$numInCat < 1000),]

Top 1000 Results

Over enriched categories (16)

rowN <- max(minRow, sum(go$q.value<=0.05))
cat(kable(go[1:rowN,c("category","term","numInCat","numDEInCat","q.value")],format="html"));

Under enriched (0)

go <- go[order(go$under_represented_pvalue),]
rowN <- max(minRow, sum(go$q.value2<=0.05))
cat(kable(go[1:rowN,c("category","term","numInCat","numDEInCat","q.value2")],format="html"));

Positive Effect

Over enriched categories (0)

rowN <- max(minRow, sum(goPos$q.value<=0.05))
cat(kable(goPos[1:rowN,c("category","term","numInCat","numDEInCat","q.value")],format="html"));

Under enriched (0)

goPos <- goPos[order(goPos$under_represented_pvalue),]
rowN <- max(minRow, sum(goPos$q.value2<=0.05))
cat(kable(goPos[1:rowN,c("category","term","numInCat","numDEInCat","q.value2")],format="html"));

Negative Effect

Over enriched categories (35)

Under enriched (0)

Final Step: csv output

write.csv(go,file=paste("csv/", outFile,"_main.csv",sep=''), row.names=FALSE)
write.csv(goPos,file=paste("csv/", outFile,"Pos.csv",sep=''), row.names=FALSE)
write.csv(goNeg,file=paste("csv/", outFile,"Neg.csv",sep=''), row.names=FALSE)