PacBio palindromic amplicons

[drkbdc]

Hi,

We are trying to perform PacBio sequencing on 2-4 kb amplicons that have palindromic sequence. My understanding is that when building the circular consensus, the CCS Reads that are palindromic are flagged and discarded since it identifies the reads as misidentification of the adaptor sequence. Since our amplicons are predominantly palindromic, is there a way to silence this feature in the ConsensusTools' code?

Thank you.

[ndelaney]

Yes, you are correct that CCS by default flags and discards palindromic sequences. If using the latest version of SMRTAnalysis, you can disable this feature by passing in the argument:

--disablePalindromeFilter

This option was only added recently, so to verify your software includes it you can run:

ConsensusTools.sh help CircularConsensus

And you should see it in the list of options.

To explain the palindrome filter: This is designed to filter out cases where an adapter is "missed" in the informatics pipeline, so that the read contains the forward strand, the adapter and the reverse strand when it should really have been divided up into two subreads. Missed adapters are pretty uncommon (particularly with the new chemistries), but as a guard against them at the end of CCS every read with > 2 passes is aligned to it's reverse complement and if it has > 85% accuracy in this alignment, it will be excluded by this filter. The total number of reads removed by this filter is printed in the output report, and you can see if it is affecting your sample by examining that.

[drkbdc]

Thank you for the quick response. Our sequencing core runs v2.3 of the SMRT Analysis tools. If by recent, you are referring to v3.0, we may be out of luck...

We know that palindromic sequences are problematic since more than 36,000 reads identified as 'CCS Read was palindrome' (22.61 %) were discarded. I will try your suggestion.

[ndelaney]

The patch was for the 2.3 version (in particular the second "p2" patch), so you hopefully have a recent enough version (I believe the option was implemented at the same time as the reporting feature, so I suspect it is there).

[drkbdc]

So indeed our build has the option of disabling palindrome filtering. I've been looking at the structure of the SMRT analysis tools, and it seems the ConsensusTools.sh script is ultimately wrapped by the smrtpipe.py python script that takes xml files (settings.xml and input.xml) as arguments, which in turn define the specific protocols/modules and files of reads to be ran by the pipeline.

The only thing I am a bit puzzled about is how exactly to invoke the --disablePalindromeFilter option through the --params option that goes with smrtpipe.py.

Suggestions?

[rhall]

Unfortunately unless the parameter is defined in the code it is not possible to pass arbitrary variables to tasks when using smrtpipe.py. If you are comfortable making some python code changes, I can probably find the files that would need to be hacked.

[ndelaney]

Unfortunately, I don't think it is possible to use that option through pbsmrtpipe. I would just run ConsensusTools.sh directly through the command line. i.e.

ConsensusTools.sh CircularConsensus --disablePalindromeFilter --fofn=input.fofn ... etc.

[drkbdc]

Ok. Thank you. I will give it a shot.