use pgenlib from pypi (resolves #16)

docs/api/data.rst

@@ -156,13 +156,6 @@ The :class:`GenotypesPLINK` class offers experimental support for reading and wr
 
 	The time required to load various genotype file formats.
 
-.. warning::
-	Use of this class is not officially supported yet because it relies upon the as-yet-unpublished ``pgenlib`` python library. See `issue #16 <https://github.com/gymrek-lab/haptools/pull/16>`_ for current progress on this challenge. In the meantime, you must install the library manually from Github via ``pip``.
-
-	.. code-block:: bash
-
-		pip install git+https://github.com/chrchang/plink-ng.git#subdirectory=2.0/Python
-
 The :class:`GenotypesPLINK` class inherits from the :class:`GenotypesRefAlt` class, so it has all the same methods and properties. Loading genotypes is the exact same, for example.
 
 .. code-block:: python

docs/formats/genotypes.rst

@@ -4,19 +4,20 @@
 Genotypes
 =========
 
-Genotype files must be specified as VCF or BCF files.
+.. figure:: https://drive.google.com/uc?export=view&id=1_JARKJQ0LX-DzL0XsHW1aiQgLCOJ1ZvC
 
-.. _formats-genotypesplink:
+	The time required to load various genotype file formats.
 
-There is also experimental support for `PLINK2 PGEN <https://github.com/chrchang/plink-ng/blob/master/pgen_spec/pgen_spec.pdf>`_ files in some commands. These files can be loaded much more quickly than VCFs, so we highly recommend using them if you're working with large datasets. See the documentation for the :class:`GenotypesPLINK` class in :ref:`the API docs <api-data-genotypesplink>` for more information.
+VCF/BCF
+-------
 
-.. figure:: https://drive.google.com/uc?export=view&id=1_JARKJQ0LX-DzL0XsHW1aiQgLCOJ1ZvC
+Genotype files must be specified as VCF or BCF files. They can be bgzip-compressed.
 
-	The time required to load various genotype file formats.
+.. _formats-genotypesplink:
 
-.. warning::
-	PGEN files are not officially supported yet because our code relies upon the as-yet-unpublished ``pgenlib`` python library. See `issue #16 <https://github.com/gymrek-lab/haptools/pull/16>`_ for current progress on this challenge. In the meantime, you must install the library manually from Github via ``pip``.
+PLINK2 PGEN
+-----------
 
-	.. code-block:: bash
+There is also experimental support for `PLINK2 PGEN <https://github.com/chrchang/plink-ng/blob/master/pgen_spec/pgen_spec.pdf>`_ files in some commands. These files can be loaded much more quickly than VCFs, so we highly recommend using them if you're working with large datasets. See the documentation for the :class:`GenotypesPLINK` class in :ref:`the API docs <api-data-genotypesplink>` for more information.
 
-		pip install git+https://github.com/chrchang/plink-ng.git#subdirectory=2.0/Python
+If you run out memory when using PGEN files, consider reading variants from the file in chunks via the ``--chunk-size`` parameter.

haptools/data/genotypes.py

@@ -1,6 +1,5 @@
 from __future__ import annotations
 import re
-import subprocess
 from csv import reader
 from pathlib import Path
 from typing import Iterator
@@ -10,6 +9,7 @@
 
 import numpy as np
 import numpy.typing as npt
+from pgenlib import PgenReader, PgenWriter
 from cyvcf2 import VCF, Variant
 from pysam import VariantFile, TabixFile
 
@@ -933,16 +933,6 @@ def read(
             See documentation for :py:attr:`~.GenotypesRefAlt.read`
         """
         super(Genotypes, self).read()
-        # TODO: figure out how to install this package
-        try:
-            from pgenlib import PgenReader
-        except ImportError:
-            raise ImportError(
-                "We cannot read PGEN files without the pgenlib library. Please install"
-                " pgenlib via\npip install git+https://github.com/chrchang/plink-ng.gi"
-                "t#subdirectory=2.0/Python"
-            )
-
         sample_idxs = self.read_samples(samples)
         with PgenReader(
             bytes(str(self.fname), "utf8"), sample_subset=sample_idxs
@@ -1097,16 +1087,6 @@ def __iter__(
             See documentation for :py:meth:`~.GenotypesPLINK._iterate`
         """
         super(Genotypes, self).read()
-        # TODO: figure out how to install this package
-        try:
-            from pgenlib import PgenReader
-        except ImportError:
-            raise ImportError(
-                "We cannot read PGEN files without the pgenlib library. Please install"
-                " pgenlib via\npip install git+https://github.com/chrchang/plink-ng.gi"
-                "t#subdirectory=2.0/Python"
-            )
-
         sample_idxs = self.read_samples(samples)
         pgen = PgenReader(bytes(str(self.fname), "utf8"), sample_subset=sample_idxs)
         # call another function to force the lines above to be run immediately
@@ -1162,17 +1142,6 @@ def write(self):
         # write the psam and pvar files
         self.write_samples()
         self.write_variants()
-
-        # TODO: figure out how to install this package
-        try:
-            from pgenlib import PgenWriter
-        except ImportError:
-            raise ImportError(
-                "We cannot write PGEN files without the pgenlib library. Please "
-                "install pgenlib via\npip install git+https://github.com/chrchang/plin"
-                "k-ng.git#subdirectory=2.0/Python"
-            )
-
         # transpose the data b/c pgenwriter expects things in "variant-major" order
         # (ie where variants are rows instead of samples)
         data = self.data.transpose((1, 0, 2))

pyproject.toml

@@ -15,6 +15,7 @@ pysam = "^0.19.1"
 cyvcf2 = "^0.30.14"
 brewer2mpl = "^1.4.1"
 matplotlib = "^3.5.1"
+Pgenlib = "^0.81.2"
 
 # docs
 # these belong in dev-dependencies, but RTD doesn't support that yet -- see

tests/test_data.py

@@ -252,7 +252,6 @@ def test_subset_genotypes(self):
 
 class TestGenotypesPLINK:
     def _get_fake_genotypes_plink(self):
-        pgenlib = pytest.importorskip("pgenlib")
         gts_ref_alt = TestGenotypes()._get_fake_genotypes_refalt()
         gts = GenotypesPLINK(gts_ref_alt.fname)
         gts.data = gts_ref_alt.data