From c7181acc88cdd8e971e4cd71d01582ae244f00d9 Mon Sep 17 00:00:00 2001
From: Muhammed Hasan Celik <muhammedhasancelik@gmail.com>
Date: Wed, 8 Jan 2020 16:59:57 +0100
Subject: [PATCH] compare chroms, resolves #55

---
 kipoiseq/utils.py          |  62 +++++++++++++++++++++++++++++++++++--
 setup.py                   |   3 +-
 tests/data/test.vcf.gz     | Bin 573 -> 577 bytes
 tests/data/test.vcf.gz.tbi | Bin 106 -> 106 bytes
 tests/test_3_utils.py      |  36 ++++++++++++++++++++-
 5 files changed, 97 insertions(+), 4 deletions(-)

diff --git a/kipoiseq/utils.py b/kipoiseq/utils.py
index 5610c37..5d97703 100644
--- a/kipoiseq/utils.py
+++ b/kipoiseq/utils.py
@@ -1,5 +1,7 @@
-import numpy as np
 from six import string_types
+import numpy as np
+import pyranges
+from kipoiseq.extractors import MultiSampleVCF, FastaStringExtractor
 
 
 # alphabets:
@@ -34,6 +36,62 @@ def parse_dtype(dtype):
         try:
             return eval(dtype)
         except Exception as e:
-            raise ValueError("Unable to parse dtype: {}. \nException: {}".format(dtype, e))
+            raise ValueError(
+                "Unable to parse dtype: {}. \nException: {}".format(dtype, e))
     else:
         return dtype
+
+
+def _get_chrom_annotation(source):
+    if type(source) == FastaStringExtractor:
+        return set(source.fasta.keys())
+    elif type(source) == MultiSampleVCF:
+        return set(source.seqnames)
+    elif type(source) == pyranges.PyRanges:
+        return set(source.Chromosome)
+    else:
+        raise ValueError('source `%s` is not valid is not valid because '
+                         ' source type `%s` is not supported.'
+                         % (repr(source), type(source)))
+
+
+def compare_chrom_annotation(sources, strategy='some', core_chroms=None):
+    """Compares chromosome annotations from different sources.
+        Throws exception iif annotations are not compatible.
+
+    # Arguments:
+        sources: list of different objects. vcf, fasta, pyranges are valid.
+        strategy: comparison strategy. `some` means some intersection excepted
+          or  `all` all chromosomes should be same.
+        core_chroms: chromosomes must exist.
+
+    # Returns:
+        chroms common cross files.
+
+    # Example:
+        ```python
+          >>> sources = [
+                MultiSampleVCF(...),
+                FastaStringExtractor(...),
+                pyranges,
+                pyranges,
+                MultiSampleVCF(...)
+              ]
+          >>> compare_chrom_annotation(sources, strategy='all')
+        ```
+    """
+    if not len(sources) > 1:
+        raise ValueError(
+            'At least two item should gived as sources to compare')
+
+    chroms = list(map(_get_chrom_annotation, sources))
+
+    if strategy == 'all':
+        assert all(chroms[0] == i for i in chroms), \
+            'chroms annotations are not all same.'
+        return chroms[0]
+    elif strategy == 'some':
+        chrom_intersect = set.intersection(*chroms)
+        assert len(chrom_intersect) > 0, \
+            'there is not intersection between chromosomes.'
+        return chrom_intersect
diff --git a/setup.py b/setup.py
index 4779d5e..59ff134 100755
--- a/setup.py
+++ b/setup.py
@@ -26,6 +26,7 @@
     "jedi",
     "pytest>=3.3.1",
     "pytest-xdist",  # running tests in parallel
+    "pytest-mock",
     "pytest-pep8",  # see https://github.com/kipoi/kipoi/issues/91
     "pytest-cov",
     "coveralls",
@@ -36,7 +37,7 @@
     # "genomelake",
     "keras",
     "tensorflow",
-    "pybedtools"
+    "pybedtools",
 ]
 
 setup(
diff --git a/tests/data/test.vcf.gz b/tests/data/test.vcf.gz
index d96030e736f3c1f0c1a78351977ff7ccb8e02699..e37d8495bb1302350865c635e79b7e8cee0d7439 100644
GIT binary patch
delta 559
zcmV+~0?_@v1i=J<ABzYC000000RIL6LPG)oBm#|;U2mH(6o#+eUtu-c9io^vm0FES
zER-ZtLcZ8`?Z8Q}<QR&bPLcNGXFd`Icgo;~Bm2D1d3}6fFxW8;ySzM5<^3FmA5XS(
zHy9Kdl}zltf|iO=u2+&=D7N>$P6IE=N@8&kdzE?Q+4-h_cc)RfsQ03c6&(*-C_Q4u
zrvkjGP_T#63P2_$D-_Fxci*Joe1|vmP;ij=7gHGysG|2BoFDbU#kry!iwl%;4i%IM
zoK$g&_aJhm_0qtP&UpI<-mh^u$&W%+%nQuIx^Wn2l3{-qw`Hv=MP+~`Jpj(jRC}Tl
zb0!s}-SWYIvj6D<C{$*+4-NMr&#C&S_ppTbBZ%Z=5PZYHMwR;74TMNQs#1GsV8$7~
zpApX41T_HZb`ve}0jeySvBJXJBjYJ1?n{?7asSJj&|*#t_li~{!s#r2Ti1|x;&dge
zBjrq;@D_Of8xzMA$e|&aft30NslY?$OEpij9ny?{egIC-LmDpDb3d*T<G4p3$L-jB
z60iD9!C8&l<Cv6wv$O`9UX9X~esXc)-q7pz^G`NLll5YbmWvIV259{jBK<L-)enD$
znl7X2f>BcoMZTjS(!SmH{diaXRL~Q$QH)0D#eUY2jmGioWQFVp`*ldHAwkX`r;W%1
xX`)6ZIMm5L8vBSiSHdb?PlI0TMvqoo&#iU-XnaFjRq9Ir0#<<yQ#6wv0w0k&4*&oF

literal 573
zcmV-D0>b?tiwFb&00000{{{d;LjnLG0*#YhZ<{a_hOgaUVKv$vqL?<7T8&98lq6C@
zK49Cm11G_fV<>hyMcR*_`A8JpDGN6one#s9wf!;}Y#E1bQSGVnevZPAC)>Fj49c8J
zCU#y)E5#_+D`_DV+j(E7fk!M3VyAMCJb%+)rct<P)}z>p4*LyMx+6K3;7x^s9aL5T
zGObvtSRuUowgnek9MgTtLE>Lb<v5^<-uG~RbPFyi6y;c)p_Fr|p-kYkt~0y?Q7EmK
z27YwH+gI>@gTrZY5UOULVHVbhhJmIz_Gd{q)~Z`n4p`DX;G)X3Cn_;#QbE=qA1rU*
zT>z!Z4fmnpKI8>e|MVVK@O}W19u0zT7>HG+uiQY06r`%OhZbh+;rj{UoJ>#ykZs~<
ziT6-<$()rI-W(XuFmYe{tcm+y)`XUGTDljs1`$qY$=kYtyc4G{Ssf^6>WDYMi{F?y
zra%r2!3?C-H%J8@I$!EM$u6Wh{Q#VuLK-gCb3bVi<K%{O9Cxw#BwpQ;f|D9|+c7D(
zWoZpGyNJ?-esXr<UeW90=bvnhChNr<Ef+DG259{jBK<R<)enD$+EYgL1*3KrihM^u
zq<g#F`|&Q#D}tVojS@6MFZQ#pY&1?@Co5z>*snuk4GD7oI9){^NL#hSp>Fok*hj><
z5Z3KF4SKC#JzCv7x6%2d@fB%3sW1Ht-@6KFFa-brABzYC000000RIL6LPG)o8vp|U
L0000000000=`{Qs

diff --git a/tests/data/test.vcf.gz.tbi b/tests/data/test.vcf.gz.tbi
index 303734c6182ddd2dd3bd7d460b7e875accd5632c..b385012cf090c177f01988d3925f401248d8a393 100644
GIT binary patch
delta 35
rcmd1GnxHQ|S0^BnEp$WGdQXAg7mv>HtY8S$XJKGC`RDkfiCzi-_+$=t

delta 35
rcmd1GnxHSeRwp2m&2>Z7dQXAg7mv>HtY8S$XJKHNdPYlgqL%^y@G}g&

diff --git a/tests/test_3_utils.py b/tests/test_3_utils.py
index 7520b77..201bd3e 100644
--- a/tests/test_3_utils.py
+++ b/tests/test_3_utils.py
@@ -1,6 +1,10 @@
+from conftest import vcf_file, sample_5kb_fasta_file, example_intervals_bed
 import pytest
 import numpy as np
-from kipoiseq.utils import parse_alphabet, parse_dtype
+import pyranges
+from kipoiseq.extractors import FastaStringExtractor, MultiSampleVCF
+from kipoiseq.utils import parse_alphabet, parse_dtype, \
+    compare_chrom_annotation
 
 
 def test_parse_alphabet():
@@ -17,3 +21,33 @@ def test_parse_type():
     assert parse_dtype('float') == float
     assert parse_dtype(float) == float
     assert parse_dtype("np.float32") == np.float32
+
+
+def test_compare_chrom_annotation():
+    sources = [
+        MultiSampleVCF(vcf_file),
+        FastaStringExtractor(sample_5kb_fasta_file),
+        pyranges.read_bed(example_intervals_bed)
+    ]
+
+    with pytest.raises(ValueError):
+        assert compare_chrom_annotation([])
+
+    with pytest.raises(ValueError):
+        assert compare_chrom_annotation([object()])
+
+    assert compare_chrom_annotation(sources) == {'chr1'}
+    assert compare_chrom_annotation(sources, strategy='all') == {'chr1'}
+
+    with pytest.raises(AssertionError) as exception:
+        sources[1].fasta = {'chr1': '', 'chr2': '', 'chr3': ''}
+        compare_chrom_annotation(sources, strategy='all')
+
+    assert str(exception.value) == 'chroms annotations are not all same.'
+
+    assert compare_chrom_annotation(sources) == {'chr1'}
+
+    with pytest.raises(AssertionError) as exception:
+        sources[1].fasta = {'chr2': '', 'chr3': ''}
+        compare_chrom_annotation(sources)
+    assert str(exception.value) == 'there is not intersection between chromosomes.'