RNA-seq quality assessment

Question

[amy16]

I've completed mapping the RNA-seq to the reference genome using HISAT2. Now I would like to assess the quality of the samples. What is the recommended tool to do it? PS: I am only a beginner.

[GenoMax]

RNA-SeQC for one. What do alignment stats look like?

[amy16]

HISAT2 summary stats:

   Total pairs: 30462111
            Aligned concordantly or discordantly 0 time: 1887598 (6.20%)
            Aligned concordantly 1 time: 24214657 (79.49%)
            Aligned concordantly >1 times: 1792176 (5.88%)
            Aligned discordantly 1 time: 2567680 (8.43%)
    Total unpaired reads: 3775196
            Aligned 0 time: 1903554 (50.42%)
            Aligned 1 time: 1501779 (39.78%)
            Aligned >1 times: 369863 (9.80%)
    Overall alignment rate: 96.88%
  

Answer 1

[mforde84]

FastQC is good for both pre- and post- alignment QC. Picard tools has a bunch of QC related tools you can use. For starters, I'd first look at the % of uniquely mapped reads to get a general idea of how well your sequences aligned to the reference. For human cell lines, a good alignment will be 50%+. Paraffin embedded tissue or other similar preparation methods will likely have more degradation so I'd assume that the alignments might suffer a bit. QC will typically assess the sequencing quality (how well the sequencer sequenced what you gave it to sequence), and not necessarily the quality of the material you gave the sequencer to sequence.

[amy16]

Ok. I would like to use Picard tools. I have the gene annotation file in GFF3 format. How do I convert it to REF_FLAT (File) format?

[Arindam Ghosh]

Any possible reference for the statement "For human cell lines, a good alignment will be 50%+. ".