Li Stephens backwards and tests

[astheeggeggs]

Description

This PR adds a collection of tests to check haploid and diploid Li and Stephens for forwards, backwards, and viterbi by testing a set of implementations against each other. For the backwards algorithm we also include a tree based implementation in python which uses existing tskit python functions.

All tests: test_haplotype_and_genotype_matching.py, based on testing in test_haplotype_matching.py. All matrix based and tree based LiS python functions are included in separate python files:

Matrix based: Forwards backwards fb_haploid_samples_variants.py fb_haploid_variants_samples.py fb_diploid_samples_variants.py fb_diploid_variants_samples.py Viterbi fb_haploid_samples_variants.py fb_haploid_variants_samples.py fb_diploid_samples_variants.py fb_diploid_variants_samples.py

Tree based: Forwards backwards fb_haploid_variants_samples_tree.py

[codecov[bot]]

Codecov Report

Merging #2137 (12de47a) into main (3631396) will decrease coverage by 11.81%. The diff coverage is n/a.

@@             Coverage Diff             @@
##             main    #2137       +/-   ##
===========================================
- Coverage   93.37%   81.55%   -11.82%     
===========================================
  Files          27       27               
  Lines       25591    25453      -138     
  Branches     1163     1113       -50     
===========================================
- Hits        23895    20759     -3136     
- Misses       1666     4627     +2961     
- Partials       30       67       +37     

Flag	Coverage Δ
c-tests	92.33% <ø> (ø)
lwt-tests	89.05% <ø> (ø)
python-c-tests	72.11% <ø> (ø)
python-tests	?

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
python/tskit/cli.py	0.00% <0.00%> (-95.94%)	:arrow_down:
python/tskit/vcf.py	9.67% <0.00%> (-90.33%)	:arrow_down:
python/tskit/text_formats.py	10.25% <0.00%> (-89.75%)	:arrow_down:
python/tskit/drawing.py	10.41% <0.00%> (-89.02%)	:arrow_down:
python/tskit/combinatorics.py	13.70% <0.00%> (-85.66%)	:arrow_down:
python/tskit/stats.py	35.13% <0.00%> (-64.87%)	:arrow_down:
python/tskit/trees.py	48.34% <0.00%> (-49.71%)	:arrow_down:
python/tskit/util.py	59.56% <0.00%> (-40.44%)	:arrow_down:
python/tskit/metadata.py	79.68% <0.00%> (-19.34%)	:arrow_down:
python/tskit/tables.py	84.03% <0.00%> (-14.87%)	:arrow_down:
... and 1 more

---

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 3631396...12de47a. Read the comment docs.

[jeromekelleher]

Thanks @astheeggeggs! I'll take a proper look later.

[jeromekelleher]

We should be able to add lshmm to our requirements list now @astheeggeggs and compare against that rather that using these included files.

Do you want to update this PR or would it be better if I refactored the existing tests to use lshmm? (This may cause significant merge conflicts with your code, though)

[astheeggeggs]

I've updated my PR to to compare against lshmm, and only include tests for matrix against tree for haploid forwards and backwards algorithms.

[jeromekelleher]

Closing this as it's out of date (WIP on getting similar code in)