Read quality calculation for a base in CCS

[curious.genome]

Hi guys,

I was wondering if anyone could tell me how PacBio calculates the quality scores while generating CCS sequences.

For example, suppose a particular base has 3 passes, and so we get 3 different subreads that contain this base. Now, we'll collapse these three subreads into a single, higher quality "consensus" sequence.

Lets say in Subread 1 the quality score for the base is 10, in Subread 2 the quality score is 15 and in Subread 3 the quality score is 20.

How is the quality score assigned for this base when it shows up in the Circular Consensus Sequence ? Is it an average? Is there a probability distribution of errors during each pass that's taken into account ?

Thanks!

Ref : slide 25 from this presentation : https://speakerdeck.com/pacbio/speci...equencing-data

[curious.genome]

So I'm going to try to answer my own question. I spoke to a PacBio rep yesterday, and from what I gather, it seems like they have an HMM that models the quality of each position in the CCS based on pre-computed probabilities of (predominantly indel) errors at the different bases in the subreads that give rise to the CCS.

I am not sure of the details, I'm yet to read their papers on Quiver/HGAP.