Merge pull request #289 from apregier/downloader

Changes for release

.travis.yml

@@ -21,6 +21,7 @@ install:
   - deps='libgfortran pip nose coverage statsmodels numpy pandas=0.19.2 scipy'
   - conda create -q -n test-environment python=$TRAVIS_PYTHON_VERSION $deps
   - source activate test-environment
+  - pip install logzero
   - pip install svtyper
 
 # command to run tests

setup.py

@@ -31,7 +31,12 @@
     packages=find_packages(exclude=['tests']),
     include_package_data=True,
 
-    install_requires=['svtyper==0.7.0', 'numpy', 'scipy', 'statsmodels', 'pandas', 'setuptools'],
+    install_requires=['svtyper==0.7.1', 'numpy', 'scipy', 'statsmodels', 'pandas', 'setuptools',
+        'google-auth',
+        'google-cloud-storage',
+        'google-compute-engine',
+        'crcmod',
+        'logzero'],
     scripts=['scripts/create_coordinates'],
 
     entry_points={

svtools/afreq.py

@@ -103,6 +103,7 @@ def epilog():
 
 def add_arguments_to_parser(parser):
     parser.add_argument(metavar='<VCF>', dest='input_vcf', nargs='?', default=None, help='VCF input')
+    parser.add_argument('-t', '--tempdir', metavar='<DIR>', required=False, default=None, help='Directory for temp file downloads')
     parser.set_defaults(entry_point=run_from_args)
 
 def command_parser():
@@ -111,7 +112,7 @@ def command_parser():
     return parser
 
 def run_from_args(args):
-    with su.InputStream(args.input_vcf) as input_stream:
+    with su.InputStream(args.input_vcf, args.tempdir) as input_stream:
         updater = UpdateInfo(input_stream)
         updater.execute()
 

svtools/bedpetobed12.py

@@ -195,6 +195,7 @@ def add_arguments_to_parser(parser):
     parser.add_argument('-o', '--output', metavar='<BED12>', type=argparse.FileType('w'), default=sys.stdout, help='Output BED12 to write (default: stdout)')
     parser.add_argument('-n', '--name', metavar='<STRING>', default='BEDPE', help="The name of the track. Default is 'BEDPE'")
     parser.add_argument('-d', '--maxdist', metavar='<INT>', dest='dist', default=1000000, type=int, help='The minimum distance for drawing intrachromosomal features as if they are interchromosomal (i.e., without a line spanning the two footprints). Default is 1Mb.')
+    parser.add_argument('-t', '--tempdir', metavar='<DIR>', required=False, default=None, help='Directory for temp file downloads')
     parser.set_defaults(entry_point=run_from_args)
 
 def command_parser():
@@ -203,7 +204,7 @@ def command_parser():
     return parser
 
 def run_from_args(args):
-    with su.InputStream(args.input) as stream:
+    with su.InputStream(args.input, args.tempdir) as stream:
         processBEDPE(stream, args.name, args.dist, args.output)
 
 if __name__ == "__main__":

svtools/bedpetovcf.py

@@ -60,6 +60,7 @@ def epilog():
 def add_arguments_to_parser(parser):
     parser.add_argument('-i', '--input', metavar='<BEDPE>', default=None, help='BEDPE input (default: stdin)')
     parser.add_argument('-o', '--output', metavar='<VCF>', type=argparse.FileType('w'), default=sys.stdout, help='Output VCF to write (default: stdout)')
+    parser.add_argument('-t', '--tempdir', metavar='<DIR>', required=False, default=None, help='Directory for temp file downloads')
     parser.set_defaults(entry_point=run_from_args)
 
 def command_parser():
@@ -68,7 +69,7 @@ def command_parser():
     return parser
 
 def run_from_args(args):
-    with su.InputStream(args.input) as stream:
+    with su.InputStream(args.input, args.tempdir) as stream:
         bedpeToVcf(stream, args.output)
 
 if __name__ == '__main__':

svtools/breakpoint.py

@@ -1,3 +1,5 @@
+import sys
+
 import l_bp
 from exceptions import MissingProbabilitiesException
 

svtools/copynumber.py

@@ -102,6 +102,7 @@ def add_arguments_to_parser(parser):
     parser.add_argument('--cnvnator', metavar='<PATH>', required=True, help='path to cnvnator binary for the cnvnator used by speedseq (required)')
     parser.add_argument('-i', '--input', metavar='<VCF>', default=None, help='VCF input')
     parser.add_argument('-o', '--output', metavar='<PATH>', type=argparse.FileType('w'), default=sys.stdout, help='output VCF to write (default: stdout)')
+    parser.add_argument('-t', '--tempdir', metavar='<DIR>', required=False, default=None, help='Directory for temp file downloads')
     parser.set_defaults(entry_point=run_from_args)
 
 def command_parser():
@@ -110,7 +111,7 @@ def command_parser():
     return parser
 
 def run_from_args(args):
-    with su.InputStream(args.input) as stream:
+    with su.InputStream(args.input, args.tempdir) as stream:
         sv_readdepth(stream, args.sample, args.root, args.window, args.output, args.cnvnator, args.coordinates)
 
 # initialize the script

svtools/l_bp.py

@@ -1,5 +1,10 @@
+import sys, re, os
+
+ar=os.path.dirname(os.path.realpath(__file__)).split('/')
+svtpath='/'.join(ar[0:(len(ar)-1)])
+sys.path.insert(1, svtpath)
 from svtools.utils import parse_bnd_alt_string
-import re
+
 
 def find_all(a_str, sub):
     '''
@@ -64,8 +69,9 @@ def parse_vcf(vcf_file_stream, vcf_lines, vcf_headers, add_sname=True, include_r
                             mid=A[7][pos_s:pos_e]
                             post=A[7][pos_e:]
                             A[7] = pre + mid + ',--:0' + post
-
-                        A[7] = 'SVTYPE=INV' + A[7][10:] + ';END=' + o_pos
+                        if ';END=' not in A[7]:
+                            A[7] = A[7] + ';END=' + o_pos
+                        A[7]=A[7].replace('SVTYPE=BND', 'SVTYPE=INV')
                         A[4] = '<INV>'
                         l = '\t'.join(A) + '\n'
                 vcf_lines.append(l)
@@ -119,6 +125,8 @@ def split_v(l):
         sep, chr_r, pos_r = parse_bnd_alt_string(A[4])
         m['END'] = pos_r
         pos_r = int(pos_r)
+    elif m['SVTYPE'] == 'INS':
+        pos_r=pos_l+int(m['SVLEN'])
     else:
         pos_r = int(m['END'])
 

svtools/lmerge.py

@@ -1,3 +1,5 @@
+import sys
+
 import svtools.l_bp as l_bp
 from svtools.breakpoint import Breakpoint
 import svtools.logspace as ls
@@ -291,6 +293,8 @@ def create_merged_variant(BP, c, v_id, vcf, use_product, weighting_scheme='unwei
 
     if var.get_info('SVTYPE') == 'BND':
         var.set_info('EVENT', str(v_id))
+    elif var.get_info('SVTYPE') == 'INS':
+        var.set_info('END', new_pos_L)
     else:
         var.set_info('END', new_pos_R )
 
@@ -419,7 +423,7 @@ def write_var(var, vcf_out, include_genotypes=False):
 
         invtobnd(var)
 
-    if var.alt not in ['<DEL>', '<DUP>', '<INV>']:
+    if var.alt not in ['<DEL>', '<DUP>', '<INV>', '<INS>']:
 
         var.var_id=str(v_id)+'_1'
         var.set_info('EVENT', v_id)
@@ -537,7 +541,7 @@ def r_cluster(BP_l, sample_order, v_id, use_product, vcf, vcf_out, include_genot
 
 
 
-def l_cluster_by_line(file_name, percent_slop=0, fixed_slop=0, use_product=False, include_genotypes=False, weighting_scheme='unweighted'):
+def l_cluster_by_line(file_name, tempdir, percent_slop=0, fixed_slop=0, use_product=False, include_genotypes=False, weighting_scheme='unweighted'):
 
     v_id = 0
 
@@ -546,7 +550,7 @@ def l_cluster_by_line(file_name, percent_slop=0, fixed_slop=0, use_product=False
     vcf = Vcf()
     vcf_out=sys.stdout
 
-    with InputStream(file_name) as vcf_stream:
+    with InputStream(file_name, tempdir) as vcf_stream:
 
         BP_l = []
         BP_sv_type = ''
@@ -616,6 +620,7 @@ def add_arguments_to_parser(parser):
     parser.add_argument('--sum', dest='use_product', action='store_false', default=True, help='calculate breakpoint PDF and position using sum algorithm instead of product')
     parser.add_argument('-g', dest='include_genotypes', action='store_true', default=False, help='include original genotypes in output. When multiple variants are merged, the last will dictate the genotype field')
     parser.add_argument('-w', dest='weighting_scheme', metavar='<STRING>', default="unweighted", choices=['carrier_wt', 'evidence_wt'], help='weighting scheme (intended for use in tiered merging), options: unweighted, carrier_wt, evidence_wt')
+    parser.add_argument('-t', '--tempdir', metavar='<DIR>', required=False, default=None, help='Directory for temp file downloads')
     parser.set_defaults(entry_point=run_from_args)
 
 def command_parser():
@@ -625,6 +630,7 @@ def command_parser():
 
 def run_from_args(args):
     l_cluster_by_line(args.inFile,
+            args.tempdir,
             percent_slop=args.percent_slop,
             fixed_slop=args.fixed_slop,
             use_product=args.use_product,