How to use RSEM to quantify genes and transcript expression from STAR outpute file (Aligned.toTranscriptome.out.bam)

Question

[otieno43]

Have already run normal Star alignment and obtained Aligned.toTranscriptome.out.bam file which should be compatible with RSEM tool. However, when I use RSEM to quantify genes and transcript expression I do not any result. I think I do not get what one part rsem-calculate-expression.

If this is the command:

rsem-calculate-expression --bam --no-bam-output -p 10 \
--paired-end --forward-prob 0 \
RNASEQ_data/star_GM12878_rep1/Aligned.toTranscriptome.out.bam # (star output file)
GENOME_data/rsem/GRCh38 \ #(rsem reference genome)
RNASEQ_data/rsem_GM12878_rep1/rsem \ # What is this suppose to be? I think this is what I am missing)
>& RNASEQ_data/rsem_GM12878_rep1/rsem.log

Any help please.

Erick

[Ram]

The "what is this supposed to be" part is the output prefix. Are you using >& when you mean to use &>? In any case, use >rsem.out 2>rsem.err and show us the contents of both files.

[otieno43]

Hello Ram,

I have tried what you suggested, I still got an error. The error reads, "-tag XM" failed! Plase check if you provide correct parameters/options for the pipeline!"

rsem-calculate-expression --bam --no-bam-output -p 10 \
--paired-end --forward-prob 0 \
RNASEQ_data/star_GM12878_rep1/Aligned.toTranscriptome.out.bam \
GENOME_data/rsem/GRCh38 \
RNASEQ_data/rsem_GM12878_rep1/rsem \
> RNASEQ_data/rsem_GM12878_rep1/rsem.log 2>RNASEQ_data/rsem_GM12878_rep1/rsem.err

[otieno43]

Another error is found in rsem.err file reads, "The SAM/BAM file declares more reference sequences (25014) than RSEM knows (20519)!"

This seems strange because I used the same reference genome (fasta) and annotation (gft) file to generate rsem reference genome and star index. I used stand alone star to build index same to rsem ref genome.

[Ram]

Can you share your STAR command as well? See the STAR command line used by RSEM here: https://github.com/deweylab/RSEM/blob/master/rsem-calculate-expression#L457

You will need to dig deeper to understand why your STAR generated BAM was not compatible with RSEM. I also think that your STAR and RSEM are not being run against the same prepared index or at least not with compatible parameters.