AssembleContigs Inquiry

[bshim181]

I have been seeing some discrepancies between the percentage of reads that have full VDJ region coverage and the percentage of clones that demonstrate succesfully assembly of VDJ region contigs.
This is a nanopore data and I utilize generic-ont preset with
--assemble-contigs-by VDJRegion parameter.

In the assmebleContigs Report, I see

I expected higher percentage of clones to have successfully assembly of contigs since high percentage of reads demonstrate alignment till the VDJ region. There seems to be some percentages of clones filtered out after assemble.
This is the assemble report.

I am trying to answer based on this data, what percent of clones would have full reconstructed VDJ regions. any help would be appreciated!

[mizraelson]

Hi, hard to say without looking at the data, but knowing the sequencing quality issues with ONT, I think that during the full length assembly process many clones got filtered out because of that. The alignment step does not discard sequences based on Phred quality hence you see a high coverage rate. Then, when you assemble clones by the CDR3 sequence, you lose some reads (7.29%) due to quality. But with assembleContigs, once you start considering the full receptor length, more clones get discarded because of low quality in other regions.

[bshim181]

I see. is there a way to rescue those clones?
i was trying to modify some of the assemble parameters like this one. Is this going to rescue clones that are discarded based on low quality bp?
-Massemble.cloneAssemblerParameters.minimalQuality=1

[mizraelson]

Yes, you can use the two parameters bellow:

-Massemble.cloneAssemblerParameters.badQualityThreshold=0
-Massemble.cloneAssemblerParameters.minimalQuality=0

[bshim181]

This is what i see when I modify the minimal quality parameter. It shows a drastic increase in the number of clonotypes initially assembled, from 70,000 to 991,004, but i was wondering if there is way to rescue clones during the assembleContigs process (I might not want those extra clones that we are not maybe fully confident about. I am assuming 90% of those extra clones are singleton)

[mizraelson]

If you only need the clones you are fully confident in, it is better to leave the parameters as is. You can also try --assemble-clonotypes-by VDJRegion --remove-step assembleContigs to assemble clones by the full VDJRegion right away since you have long read sequencing.