Writing fasta files from tree sequences

[maz-mckellar]

I've written some code to cover integrating fasta output from tree sequences into tskit, following from issue #338 and have included various tests for it.
Happy to hear feedback on this (e.g. @jeromekelleher, @hyanwong).

[jeromekelleher]

Great stuff, thanks @marianne-aspbury. How about we go through this in person tomorrow?

[maz-mckellar]

Yep, sounds good, thanks

[codecov]

Codecov Report

Merging #350 into master will increase coverage by 1.11%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #350      +/-   ##
==========================================
+ Coverage   86.48%   87.59%   +1.11%     
==========================================
  Files          20       19       -1     
  Lines       14024    10275    -3749     
  Branches     2740     1890     -850     
==========================================
- Hits        12128     9000    -3128     
+ Misses        977      756     -221     
+ Partials      919      519     -400

Flag	Coverage Δ
#c_tests	87.59% <100%> (+0.02%)	⬆️
#python_c_tests	?
#python_tests	99.24% <100%> (ø)	⬆️

Impacted Files	Coverage Δ
python/tskit/cli.py	97.84% <100%> (+0.11%)	⬆️
python/tskit/trees.py	98.68% <100%> (+0.01%)	⬆️
python/_tskitmodule.c

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 74461f4...a8bddcc. Read the comment docs.

[jeromekelleher]

Looks great, thanks @marianne-aspbury. We're basically there I think.

It's probably as well to look at the CLI as well here We basically want to duplicate the VCF option in tskit/cli.py and tests/test_cli.py; give me a shout if you have questions.

[maz-mckellar]

Added functionality for printing fastas on command line, pretty much copied directly from the vcf CLI code and all its tests.

[hyanwong]

Sorry - I've not been following this. Note my reservations in #326 (comment) about making sure that FASTA files (and haplotypes in general) output non-variable sites, at a minimum with a dot, but possibly filling in via a reference genome.

[petrelharp]

I don't think we're ready to deal with invariant sites yet, @hyanwong - to do that we need to distinguish tree sequences with integer positions and a reference sequence from the current infinite-sites notions. I'm in favor of doing this, but it shouldn't get in the way of this effort.

[jeromekelleher]

I don't think we're ready to deal with invariant sites yet, @hyanwong - to do that we need to distinguish tree sequences with integer positions and a reference sequence from the current infinite-sites notions. I'm in favor of doing this, but it shouldn't get in the way of this effort.

Agreed. Let's consider what we can do as a stopgap for finite sites as part of #326 (without changing the default behaviour; we don't break people's code). We can then decide what the final semantics of the fasta function should be in #353 (tagged as a 0.2.3 release issue so we don't forget about it).

Let's get this PR finished up and merge first though.

[hyanwong]

I don't think we're ready to deal with invariant sites yet, @hyanwong

We've constructed tree sequences with a site at each integer position, but many sites without mutations. Isn't that the same as an invariant site? But I guess this might be though of as a bit of a hack.

[jeromekelleher]

Looks great, thanks @marianne-aspbury! Some very minor comments above --- we're ready to merge after these are addressed and the branch is rebased to bring it up to date.

[maz-mckellar]

Looks great, thanks @marianne-aspbury! Some very minor comments above --- we're ready to merge after these are addressed and the branch is rebased to bring it up to date.

Thanks @jeromekelleher, I've fixed these now