Which is a good coverage in RNA Seq data?

Question

[germano.sollazzo]

Hi all, I have some RNA Seq data (whole genome) from different strains and I'm looking for SNPs inside. In some exons I have a good coverage (nucleotide are mapped 1500 times) but in other positions I have low coverage ( 5-10-30 for example)

Do you know which is the minimum value to consider a SNP as a good candidate? I mean, if I have a SNP in the mutated strain compared to wild type and in that position I have 1000 reads mapping, I can trust it, but if in an other position I have only 15 reads mapping there, can I consider it too? or the reads are too less?

Do you know any papers about this?

Thanks in advance. Germano

[Ram]

In some exons I have a good coverage (nucleotide are mapped 1500 times) but in other positions I have low coverage ( 5-10-30 for example)

RNAseq captures qualitative data as well as quantitative data. If exon1 of a gene is found in 2 of its transcripts but exon2 is found in only 1, then even if both transcripts are expressed to the same extent, exon1 will be sequenced to 2X depth compared to exon2. You'll need to account for this when picking a threshold.

You may want to look for papers about experiments where variant calling is done on RNAseq data.