TranscriptomeSAM and multi-mapping

[Marianna Pauletto]

Hi all,

I'd like to map reads using the multi-mapping option (--outFilterMultimapNmax 10) and get the TranscriptomeSAM output in order to use RSEM.

As far as "multi-mapping" reads, I know that STAR is normally used for mapping to the genome, so the unique (multi-) mappers are defines as those mapping to 1 (>1) loci in the genome.

Since mapping is performed against the genome, if there is a gene which codes for 15 transcripts starting with the same exon, this os not a multi-mapping. Isn't it?

But then, whene the alignments are translated into transcript coordinates, how does STAR deal with these "multi-mapping" reads??

Thank you!

Best

Marianna

[Alexander Dobin]

Hi Marianna,

each genomic alignment will be converted to as many transcriptomic alignments as needed, there is no limit.

In your example, if a read maps to the exon that is common to 15 transcripts, it will be reported as genomic unique mapper in the Aligned.out.bam and as 15-mapper in the Aligned.toTranscriptome.bam.

Cheers

Alex

[Marianna Pauletto]

Hi Alex,

thank you for your answer.
Just another question: how does a multimapper read will be reported in the Aligned.out.bam? It will be mapped random in one position or it will be counted as mapped in all the multiple positions?

Best
Marianna

[Alexander Dobin]

Hi Marianna,

by default, all multimapping loci of a read are reported, if the number of those loci is <= --outFilterMultimapNmax (=10 by default).

If the number of loci exceeds this value, no loci are reported at all.

This can be changed by specifying --outSAMmultNmax  (int: max number of multiple alignments for a read that will be output to the SAM/BAM files) 

and --outMultimapperOrder (random order of alignments for each multi-mapper).

Cheers

Alex

[Marianna Pauletto]

Hi Alex,

sorry my question was not really clear.
I mean: when I look at the gene counts file, a multimapper read (mapping to different genes) will be assigned randomly to a single gene or to all the genes (once per gene)?

Thank you very much
Marianna

[Marianna Pauletto]

Sorry Alex

I'm a bit confused.
Let's try with an example.

I've mapped reads with the following parameters
-outFilterMultimapNmax 1 --quantMode TranscriptomeSAM GeneCounts

Now I'd like to use the aligned.toTranscriptome.out.bam files in RSEM and I'm wondering how reads mapping to multiple transcripts will be counted.
Since I set -outFilterMultimapNmax 1, here by default I will have one SAM line for each mapped read. Isn't it?

If I had set -outFilterMultimapNmax 10, I could have controlled the number of alignments (against transcripts) to be shown in the bam file with --outSAMmultNmax and their order (--outMultimapperOrder random if random or without specyfing anything if I prefer the approach defined in the manual as "not truly random"). Is this correct?

What do you mean saying "not truly random"?

Best
Marianna

[Alexander Dobin]

Hi Marianna,

for genomic alignments (i.e. output to Aligned.out.bam):

--outFilterMultimapNmax N controls the max number of loci N a read can map to in the genome (not transcriptome). If a read maps to 

>N loci, no alignments are output at all for this read. If a read map to <N loci, all aligments are output to Aligned.out.bam by default, but this can be changed with --outSAMmultNmax  (int: max number of multiple alignments for a read that will be output to the SAM/BAM files) and --outMultimapperOrder (random order of alignments for each multi-mapper).

Transcriptomic alignments are converted from genomic alignment.

The genomic alignments have to satisfy the --outFilterMultimapNmax N filter.

If a read maps to Mg<=N loci in the genome, then all these Mg genomic loci are converted to Mtr transcriptomic loci.

Each genomic alignment can convert to one or more transcriptomic alignments, so Mtr>=Mg.

Hope this clarifies it a a bit, please let me know if you have further questions.

Cheers

Alex