How to correct wrong adapter splitting in PacBio films?

[cklopp]

While processing IsoSeq data, I observed that close to 10% of the films had adapter splitting problems.
The pattern is shown in the graphic hereunder with 3 examples :



Each graph correspond to one film (ZMW) which contains 20 ordered reads.
The graph shows the read length found after adapter splitting.
The one on the top is as awaited.
The second one shows a couple of reads which have not been split in position 13.
The last shows four errors.

When films contain multiple reads (let's say over 5, which is common in IsoSeq), it should be quiet easy to compare each read length to the median and decide if there has been a problem in the adapter splitting.

[Magdoll]

Hi,

The CCS algorithm is aware of possible incorrect adapter calling issues. It won't look at subreads that severely diverges from the rest of the subreads in terms of length and quality.

What you see here is not entirely uncommon, but in most cases will be correctly dealt with by the CCS algorithm by discarding those anomaly subreads.

[cklopp]

Here you have a plot of the median length of the reads versus the length of the RoI for a library.



Most of the RoI lengths seem to match the median length of the ZMW reads . Is it possible to correct the RoI in which this is not the case?

[ndelaney]

I am assuming this plot is from the older version of CCS? CCS2 by default removes any subread that is over twice the size of the median subread (see here: https://github.com/PacificBioscience...nsensus.h#L288).