Remove bundled ancient samtools 0.1.19 and move to building against upstream htslib

[SoapZA]

This is much easier and shifts the complexity away from ShoRAH.

[SoapZA]

@DrYak any movement here? I would want this in before release 2.0.0

[DrYak]

Working on it, that's my next biggest point on my todo list. I think I should be able to have something by the end of the next week. Would that be okay for your timing ?

[SoapZA]

@DrYak how far are we here?

[DrYak]

Sorry got distracted by some problems the cancer-team neighbors were having

I'm working on it. Getting a hang on how pileup have changed between versions (the API that ShoRAH has been using has been moved out of the library and into samtools executables themselves )

[DrYak]

(Sorry for slow updates, was sick last week)

B2W ported to HTSLib and produces identical results for the few test cases I have laying around.

Bascially :

• I have moved the common parts between "scanning all chromosome" - "scanning a single r.o.i." into a single lamba.
  • bam_fetch has been moved from the library into samtools. I've re-implemented it in b2w.c based on samtool/bam.c, as recommended by samtools authors.
  • as there is no call-backs involved anymore (bam_fetch is simply a while loop), i've moved the fetch_func1 and fetch_func2 into the while body (each was called exactly once, anyway)
  • bam_plbuf_push has been moved from the library into samtools. I've re-implemented it in b2w.c based on samtool/bam_plbuf.c, as recommended by samtools authors.
  • as there are no call-backs involved anymore (bam_plbuf_push too is simply a while loop), I've slightly modified pile_func and call it directly from each while loop.
  • command line parameters are stored in a separate struct param.
  • work data is accessed through the lambda's capture, instead of being passed around through a struct.

Now moving onto pushing the changes to FIL, too.

[DrYak]

And FIL port to HTSLib has been pushed to the with_htslib branch.

Basically the same mods :

• the loops for amplicon mode and normal mode have been fused (as we are only adding columns to the input SNV, not touching frequencies or priors).
• bam_fetch and bam_plbuf_push re-implemented based on samtools, as recommended by HTSlib authors.

[DrYak]

Updated Travis CI yaml file to fetch HTSLib (from Debian Med to have a not to ancient version). Updated configure.ac to request C++11 (the antediluvian GCC 4.8.3 featured on Trust Thar used by defaults to older dialects).

Now Travis tests successfully pass.

Speaking of tests, I'm also able to replicate the result I'm getting from pre-HTSLib FIL and B2W.

I'll ask Susie for datafiles for which she was receiving bad/weird results that she suspected were due to bugs from a way to old samtools lib version.

I'm done on my side, I'll need your knowledge of Meson to make the same change as I did to the autotools/m4 side.

[SoapZA]

Please open a pull request so we can review it

[DrYak]

Here it is

As said, the code is not actually mergeable, the meson-side of things need to be done.

Also, maybe you'll request more clean-ups. (splitting into functions to make the logical steps stand out ?)

[DrYak]

Thanks for your help