Making genome-wide transcription level track from transcript level count

Question

[Sam]

I used Salmon to quantify RNA at the transcript level. Is there a way to map this data into a format that can be viewed in IGV?

A related question, i also tried another method of quantifying where i aligned the read with STAR and counted with GenomicAlignments. The counts from this method is easily double that of the count from Salmon. With some transcripts counted almost 8x to 10x more. I did a Pearson's correlation and it return as value of 0.95, so i'm guessing it would not significantly affect me if i am doing genome wide analysis. But i am still curious as to how this discrepancy came about.

Any input will be greatly appreciated

Regards

Answer 1

[caggtaagtat]

Hi,

I don't know about how to convert the data from salmon. I read, that you can generate pseudobam files with kalisto and maybe there is an tool for the output of salmon.

Concerning the STAR coverages, you get more reads, because it also counts gapped reads, which are not part of an already annotated transcript. Therefore, with salmon, where you only mapp the reads against annotated transcripts of a gene, you get a lot more reads, which also include splicing errors or not yet described alternative spicing events, where a gap can not be explained by already annotated transcript sequences. Salmon takes gapped reads into acoount, but only those who can be mapped to known transcripts. With gapped reads I mean gapped after alignment to the genome (like after STAR alignment). Thats why salmon an co are called pseudoalingers. They are fast and sufficient for well-annotated organisms. But you can't for example find new isoforms of a transcript.

[Sam]

I see, thank you very much for the explanation, it is very helpful!