James Bonfield

I'm not sure the quote is necessarily correct. It's the case that the zero length CIGARs weren't being used to update the bin linear index so the start..end ranges could...

I have some suboptimal Tikz to produce this type of plot:  I'm thinking maybe it needs to be less cluttered, or perhaps one like this and then some zoomed...

I finally found the source for this pic. I hunted high and low and then remembered it was on my home PC as viewing the PDF was much quicker there....

It looks like binary gibberish because that is what BAM is - the "B" is for binary. The SAM text is converted to binary and then gzipped (more or less)....

That's a different issue than the bug I reported above (which if I recall I fixed in #1323, but it's been a couple months since I made that PR so...

On the initial input (copied from above): ``` foo 4 * 0 0 * * 0 0 ACCCCCCCCCCCCCCCCCCCCT xYYYYYYYYYYYYYYYYYYYYz BC:Z:AGGGGGGT-CGGGGGGT QT:Z:XyyyyyyZ-PqqqqqqR foo 4 * 0 0 * * 0 0...

Fixing the above issue is trivial; just remove the `b[]` array and always use `b1` (or `b2` if unset). However I'm scratching my head more now when I modify the...

Even CRAM 3.x has this limit unfortunately. It's something we fixed in the CRAM 4.0 draft, but it never really gained any traction as sadly. SAM is really the only...

Has anyone even profiled it to work out where the time is spent? Is it decoding, consensus/snp calling, or encoding the output (eg bcf+bgzf could take some time).

Alas not. We did some profiling, but the algorithm is complex and rather hard to multi-thread in the current state. It's a wish-list item basically unless anyone else wants to...