Adding a check to automagically create index for VFI and VCF files if…

… they do not exist

HISTORY.rst

@@ -3,6 +3,11 @@
 History
 -------
 
+0.2.4 (06/05/2018)
+~~~~~~~~~~~~~~~~~~
+
+* Automtically generates file index if not found
+
 0.2.3 (06/05/2018)
 ~~~~~~~~~~~~~~~~~~
 

g2gtools/__init__.py

@@ -1,3 +1,3 @@
-__version__ = '0.2.3'
+__version__ = '0.2.4'
 __author__ = 'Matthew Vincent and Kwangbom \"KB\" Choi, The Jackson Laboratory'
 __email__ = 'matt.vincent@jax.org'

g2gtools/fasta_patch.py

@@ -270,8 +270,6 @@ def process(filename_fasta, filename_vci, regions, filename_output=None, bgzip=F
             dump_fasta = True
             LOG.debug("Temporary fasta file: {}".format(filename_output))
 
-        print('hello')
-
         fasta_file = fasta.FastaFile(filename_fasta)
         vci_file = vci.VCIFile(filename_vci)
 

g2gtools/fasta_transform.py

@@ -519,7 +519,7 @@ def process(filename_fasta, filename_vci, regions, filename_output=None, bgzip=F
     :type num_processes: int
     :return: Nothing
     """
-    LOG.error("in process of fixing")
+    #LOG.error("in process of fixing")
 
 
     start = time.time()
@@ -657,8 +657,6 @@ def process(filename_fasta, filename_vci, regions, filename_output=None, bgzip=F
         args = zip(all_params)
         LOG.debug(args)
 
-
-
         pool = multiprocessing.Pool(num_processes)
         results = pool.map(wrapper, args)
 

g2gtools/g2g_utils.py

@@ -293,7 +293,7 @@ def bgzip_file(original_file, new_file, delete_original=False, force=True):
         delete_file(original_file)
 
 
-def index_file(original_file, file_format="vcf"):
+def has_index_file(original_file, file_format=None):
     """
 
     :param original_file:
@@ -302,15 +302,49 @@ def index_file(original_file, file_format="vcf"):
     :return:
     """
 
+    if not file_format:
+        # try to guess the file format
+        if original_file.lower().endswith(".fa") or original_file.lower().endswith(".fasta"):
+            file_format = 'fa'
+        elif original_file.lower().endswith(".vcf"):
+            file_format = 'vcf'
+        elif original_file.lower().endswith(".vci"):
+            file_format = 'vci'
+        else:
+            raise G2GValueError("Cannot determine file format")
+
     if file_format.lower() == 'fa':
-        pysam.faidx(original_file)
+        ext = 'fai'
     elif file_format.lower() == 'vcf':
-        pysam.tabix_index(original_file, preset="vcf", force=True)
+        ext = 'tbi'
     elif file_format.lower() == 'vci':
-        pysam.tabix_index(original_file, seq_col=0, start_col=1, end_col=1, force=True)
+        ext = 'tbi'
     else:
         raise G2GValueError("Unknown file format: {0}".format(file_format))
 
+    idx_file = '{}.{}'.format(original_file, ext)
+
+    return os.path.exists(idx_file)
+
+
+def index_file(original_file, file_format="vcf", overwrite=False):
+    """
+
+    :param original_file:
+    :param new_file:
+    :param file_format:
+    :return:
+    """
+    if not overwrite or not has_index_file(original_file, file_format=file_format):
+        if file_format.lower() == 'fa':
+            pysam.faidx(original_file)
+        elif file_format.lower() == 'vcf':
+            pysam.tabix_index(original_file, preset="vcf", force=True)
+        elif file_format.lower() == 'vci':
+            pysam.tabix_index(original_file, seq_col=0, start_col=1, end_col=1, force=True)
+        else:
+            raise G2GValueError("Unknown file format: {0}".format(file_format))
+
 
 def bgzip_and_index_file(original_file, new_file, delete_original=False, force=True, file_format="vcf"):
     bgzip_file(original_file, new_file, delete_original, force)

g2gtools/vcf2vci.py

@@ -436,6 +436,8 @@ def process(vcf_files, fasta_file, output_file, strain, vcf_keep=False, passed=F
         for file_name in vcf_files:
             vcf_file = g2g_utils.check_file(file_name)
             LOG.info("VCF file: {0}".format(vcf_file))
+            LOG.info("Checking for index file, creating if needed...")
+            g2g_utils.index_file(original_file=vcf_file, file_format="vcf", overwrite=False)
 
             vcf_discard_file = None
             if vcf_keep:

g2gtools/vci.py

@@ -165,6 +165,8 @@ def __init__(self, filename, mode='r', parser=None, index=None, encoding="ascii"
 
         self._tabix_file = pysam.TabixFile(self.filename, mode=mode, parser=parser, index=index, encoding=encoding)
 
+        g2g_utils.index_file(original_file=filename, file_format="vci", overwrite=False)
+
         self.parse_header()
 
     def __getattr__(self, name):

setup.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.2.3
+current_version = 0.2.4
 commit = True
 tag = True
 

setup.py

@@ -30,7 +30,7 @@
 
 setup(
     name='g2gtools',
-    version='0.2.3',
+    version='0.2.4',
     description="A suite of tools for the reconstruction of personal diploid genomes and better coordinate conversion",
     long_description=readme + '\n\n' + history,
     author='Matthew J. Vincent and Kwangbom "KB" Choi, The Jackson Laboratory',