getDiffExpression.pl in HOMER doesn't seem be working... What am I doing wrong?

I'll start with some context:

Performed RNA-seq on two samples, one treated with 5-aza and the other with the vehicle only.
Trimmed raw fastq reads, sorted them for rRNA and tRNAs. (cufflinks/sortmerna)
Created an hg19 index, fasta files obtained from gencode (STAR)
Aligned reads to the genome (STAR)
Created tag directories with the "-keepOne" flag as suggested in HOMER's guidelines (makeTagDirectory)
Analyzed repeats with the "-noadj" flag as suggested in HOMER's guidelines analyzeRepeats.pl)
Used getDiffExpression.pl: "getDiffExpression.pl ../un_v_aza.txt untreated untreated aza aza -repeats -edgeR -log2fold 2.0 > ./un_v_aza_de.txt"

And this is the output I get:

    Differential Expression Program: edgeR
            Treating input as file generated by analyzeRepeats.pl (-repeats)
            Using edgeR to calculate differential expression/enrichment...
            Output Stats untreated vs. aza:

            Total Genes: 1397
                    Total Up-regulated in untreated vs. aza: 0 (0.000%) [log2fold>2.0, FDR<0.05]
                    Total Dn-regulated in untreated vs. aza: 0 (0.000%) [log2fold<-2.0, FDR<0.05]

I sincerely doubt there are no upregulated/downregulated repeats differences between the two samples... what am I doing wrong ?

edit: formatting

homer rna-seq repeats 5-aza rna

• 2,977 views • link

updated 9.3 years ago by k.balaji93 •

• written 9.5 years ago by ron.eitan •

1 answer

Hi, does changing the log2fold cut-off to 1 change the output? Maybe 2 is too stringent a cut-off, since you have only 1397 genes in total (seems too low). You may not have enough statistical power. I don't think there is an industry standard (sadly) for a log2fold cut-off, for a gene to be considered differentially expressed. So even setting log2fold to 0 should be technically correct.

• 0 views • link

written 9.3 years ago by k.balaji93 •