GeneCounts result has less number of genes?

[Abdul Kadir Mukarram]

Dear all,

I am trying to utilise the --quantMode GeneCounts flag in the STAR version 2.5.1b and indeed I got ReadsPerGene.out.tab file with similar counts compared to the result of another package (featureCounts from Rsubread, for example). However, I got less number of genes in my ReadsPerGene.out.tab output (around 600 less).

I supplied the gtf file in the index generation of STAR and compared to the exact same gtf file coupled with the BAM file generated by the same STAR run in other packages (which yielded the same number of genes as if I extract "gene" lines from the gtf file). My reads are paired-ended.

This was my command:

STAR --genomeDir star_genome --runThreadN 16 --genomeLoad NoSharedMemory --outSAMtype BAM SortedByCoordinate --readFilesIn /proj/R1.fastq.gz /proj/R2.fastq.gz --outSAMunmapped Within --outSAMattributes NH HI AS NM MD --outSAMstrandField intronMotif --readFilesCommand zcat --outWigStrand Stranded --outWigType bedGraph --quantMode GeneCounts

Anything I am missing here?

Many thanks in advance!

Cheers,

Kadir

[Alexander Dobin]

Hi Kadir,

STAR only looks at the "exon" lines of the gtf, not the "gene" lines. If there are any extra "genes" with no exons in the file, it could explain the difference.

If you send me a link to the gtf file you used, I can check that.

Cheers

Alex

[Luca Cozzuto]

Dear Alexander,

I have the same problem, and I checked that the missing genes do contain exons in the GTF. 

I had a look at the geneInfo.tab generated after making the index and I see that those ids are not there. Could you help me, please?

Luca

[Alexander Dobin]

Hi Luca,

if you have genes that do not have corresponding "exon" lines, you would have to make them manually.

Bascially, you can replace "gene" with "exon" in the 3rd field, and add 

transcript_id "SomethingUnique";

at the end of the line.

Cheers

Alex

[Nikelle Petrillo]

Hi Alex, 

Are multimappers included in the Genecounts option? Or only unique mappers?

Thanks!

[Alexander Dobin]

Hi Nikelle,

only unique mappers are included in the genic counts. 

Cheers

Alex