feat: a new GenotypesPLINKTR class for reading TRs from PGEN files

[aryarm]

resolves #73 (comment)

This code was primarily authored by our summer high school student, @gonzalogc1, but he had to leave before we could finish it off. Thanks for all of the help this summer, Gonzalo!

Overview

This PR enables reading and writing multiallelic variants from a PGEN file. It also adds a new GenotypesPLINKTR class capable of reading TRs from PGEN files similar to the GenotypesTR class.

Note: A newer version of Pgenlib is required as a result of this change. The pyproject.toml file has been updated accordingly.

Usage and Documentation

The interface for the GenotypesPLINK class has not changed. It will work the same as before but will cease to raise an error for multiallelic variants. This functionality also carries over to the simgenotype command.

The GenotypesPLINKTR class works as expected; it has the same methods as the GenotypesTR and GenotypesPLINK classes, but a few of them from the GenotypesPLINK class have been disabled. Refer to the documentation for a full list of the disabled methods and a more thorough description of how to use the new class.

Tests

We added new tests and a new test data file for the GenotypesPLINK and GenotypesPLINKTR classes. The tests/data/simple-tr.pgen set of files are similar to the tests/data/simple_tr.pgen set of files except that some alleles have been removed so that the VCF file passes the checks implemented by plink2 when converting from VCF to PGEN.

Future work

• feat: a dosage command for storing TR motif counts in PGEN files #221
• Try to improve TR loading speed by using TRHarmonizer objects in GenotypesPLINKTR.read_variants(). Those objects can then be reused by GenotypesPLINKTR.read(). This will require a larger refactor that is out of scope for this PR.
• now that wheels are being published for Pgenlib, we can make it a required (instead of an optional) dependency of haptools

[mlamkin7]

All good.