We sequenced a fungal genome using PacBio Sequel for a de novo assembly we want to perform. We received the subreads in BAM and FASTA format, as well as a XML file, for 2 SMRT cells (_cell1 and _cell2). Prior to combining them, I wanted to obtain CSS from the subreads. I installed pbccs and used the ccs command on both BAM files, with default parameters. However, the vast majority of my ZMWs are failing filtering. For reference, cell 1 had 11568 and cell 2 had 15542, and I am keeping only 57 and 58. Fail reasons are mostly "Lacking full passes" and "Draft above --max-length". Default options for those two parameters are --min-passes 3 and --max-length 5000.
I tried again relaxing these parameters, --min-passes 1 and --max-length 15000, and now I'm keeping 2652 and 1876, which is still low (23 and 12% of input ZMWs) but at least we have something to work with. I'm still getting a lot of "Lacking full passes" fails, but I assume I don't want CCS based on subreads that are not even complete with 1 copy.
I have some questions on this. Do default values normally work well? Does it make sense to reduce the minimum full subreads required to 1, given that subreads do not have actual Q-values associated? Regarding maximum length, would we expect CCS that long (15000) to be actual biological data, or are they likely to be artifacts? And the most relevant question, should I simply stick to the subreads for the assembly and forget about CCS, given my results?
Thank you in advance,
Sergio
I tried again relaxing these parameters, --min-passes 1 and --max-length 15000, and now I'm keeping 2652 and 1876, which is still low (23 and 12% of input ZMWs) but at least we have something to work with. I'm still getting a lot of "Lacking full passes" fails, but I assume I don't want CCS based on subreads that are not even complete with 1 copy.
I have some questions on this. Do default values normally work well? Does it make sense to reduce the minimum full subreads required to 1, given that subreads do not have actual Q-values associated? Regarding maximum length, would we expect CCS that long (15000) to be actual biological data, or are they likely to be artifacts? And the most relevant question, should I simply stick to the subreads for the assembly and forget about CCS, given my results?
Thank you in advance,
Sergio