Does Aligned.toTranscriptome.out.bam need to be sorted for RSEM input?

[Ning Li]

Hi,

I am new to use STAR and RSEM together for RNAseq alignment and quantification. I have searched a lot to find out how to make STAR and RSEM to work together. Finally, I am able to make it work. 

However, one thing I am not sure. As mentioned in ENCODE, Aligned.toTranscriptome.out.bam needs to be sorted using samtools, and then as input for RSEM. Does anyone know is it necessary to sort Aligned.toTranscriptome.out.bam for running RSEM to calculate counts? Does it affected the final results?  

I tried to test the differences of using sorted transcriptome bam vs directly transcriptome bam for running RSEM. However, the sorting step took so long time, and I couldn't run it.

Any suggestions will be appreciated. It is a question about STAR and RSEM both, so I don't know where will be proper place to post it. Since I find STAR group is the most helpful place, so give it a try.

Thank you very much for your help.

Best

Ning

[Alexander Dobin]

Hi Ning,

the Aligned.toTranscriptome.out.bam does *not* have to be sorted for RSEM.

However, if you run STAR several time on the same sample, the order of the reads in the the file will be different.

RSEM results are very slightly dependent on the order of the reads in the file. To ensure perfect reproducibility of our pipeline, we implemented re-sorting of the file for ENCODE.

However, personally I do not believe this is necessary at all.

Cheers

Alex