The PacBio Revio Throws Away Most Reads?

Summary: As far as I can tell the plot below is correct. The vast majority of ZMWs on the Revio produce less than 1 complete subread (but do produce data) and are discarded.

I’ve been thinking about this plot:

Which I created based on a few assumption in a previous post. The problem is that the Revio throwing out 45% of reads seems like “a lot”… and it niggles me a bit. What exactly is going on here… why are these reads thrown out?

PacBio provide a “fail” BAM file. But from what I can tell this only contains a subset of the failing reads, not “one representative read per ZMW”. In one example I looked at, the failed BAM contained ~350,000 reads. Where the plot above would be suggesting more like 12M.

As we shall see, the data we need lies elsewhere in the PacBio run folder…