check rna-seq alignment quality using IGV

Question

[Tonkatsu]

I have imported my bam and index files into IGV and want to check the quality of my alignments for a particular gene to ensure the reported rpkm is reasonable. However I am having trouble interpreting what I am seeing from the viewer and what I should be looking for?

[Ram]

Do you understand RPKM well enough to be able to eyeball its accuracy using IGV alignment views?

Please note that RPKM is super outdated and you shouldn't be using it anyway.

[Tonkatsu]

Thank you for the reply, I understand the basis of rpkm expression, but I guess I am having trouble with the interface of igv. All I want to do is to discriminate good vs bad quality reads and see if I can more or less be confident in the aligner/ reads observed.

[Ram]

You should be able to set a threshold - IGV will filter out reads that fail that threshold. Read the manual and play around with IGV - that's the best way to learn.

[Tonkatsu]

thanks, also just a general question if I am looking at a specific gene that is comprised of 3 exons or 2 protein coding regions, and I find that some of my reads being aligned are very small proportionally to to the entire protein coding region and located only in one of those protein coding regions. Should I consider this a "bad quality" alignment generally speaking? Similarly if the read spans the entirety of one protein coding region, but is largely absent in the other (1/2), how should I classify these alignments?

[Ram]

I'm sure RNAseq experts can answer these questions better - I'm not sure how spliced alignments are counted in quantification operations, where an exon/CDS can be used in multiple transcripts.

Why do you think those reads are bad quality? Do these reads have low pre-alignment QUAL values or low MAPQ values after the "very small" aligned part?