polishing assembled genome to QV50 value

Question

I am trying to assemble some genomes from pacbio reads using HGAP3 followed by quiver polishing. However, even after multiple rounds of quiver polishing, I cannot attain quality value 50 (QV50) for a genome.

QV50 seems to be recommended and I am able to get upto QV30 only. I have done polishing for almost 15 times already. I have tried polishing with different parameter changes like subread length cutoff and blasr maximum divergence.

Can anyone suggest me on how I can achieve an assembled genome of QV50 value?

Thank you

Answer 1

15 times is too many times. If you were going to get quiver to get you good QVs it would have happened after 2-3 rounds probably.

Some questions:

1. Why are you using quiver? Are your reads from RSII chemistry? Arrow is recommended for Sequel reads.
2. What is your dataset coverage? If you do not have very high coverage (I think >30X is usually a minimum for a eukaryotic genome) it may be that you are just not going to get exactly the values you want because assembling low coverage data is hard.
3. Have you tried other polishing tools? They might be able to get you better value. I am not an expert but some people like to use Racon or Apollo.
4. While I'm not an expert, looking into how the QVs are calculated can be helpful. I looked at the quiver docs and it looks like if the reads map ambiguously then QV is low. So if you for example have lots of haplotigs, then mapping might be ambiguous for a lot of reads, and therefore this will depress your QV values. This is therefore a problem that can come from the starting assembly and not just the polishing. You may want to go back and take another look at your starting assembly.