Merge pull request #73 from lonelyjoeparker/master

index function and file I/O exception handling

poretools/Fast5File.py

@@ -211,7 +211,7 @@ def open(self):
 			self.hdf5file = h5py.File(self.filename, 'r')
 			return True
 		except Exception, e:
-			logger.warning("Cannot open file: %s. Perhaps it is corrupt? Moving on.\n" % self.filename)
+			logger.warning("Exception:Cannot open file: %s. Perhaps it is corrupt? Moving on.\n" % self.filename)
 			return False
 
 	def guess_version(self):
@@ -222,12 +222,20 @@ def guess_version(self):
 			self.hdf5file["/Analyses/Basecall_2D_%03d/BaseCalled_template" % (self.group)]
 			return 'classic'
 		except KeyError:
+			logger.warning("KeyError:Cannot open file: %s. Perhaps it is corrupt? Moving on.\n" % self.filename)
+			pass
+		except AttributeError:
+			logger.warning("AttributeError:Cannot open file: %s. Perhaps it is corrupt? Moving on.\n" % self.filename)
 			pass
 
 		try:
 			self.hdf5file["/Analyses/Basecall_1D_%03d/BaseCalled_template" % (self.group)]
 			return 'metrichor1.16'
 		except KeyError:
+			logger.warning("KeyError:Cannot open file: %s. Perhaps it is corrupt? Moving on.\n" % self.filename)
+			pass
+		except AttributeError:
+			logger.warning("AttributeError:Cannot open file: %s. Perhaps it is corrupt? Moving on.\n" % self.filename)
 			pass
 
 		return 'prebasecalled'

poretools/index.py

@@ -0,0 +1,66 @@
+import Fast5File
+import datetime
+
+############
+#
+#	index
+#
+#   A tool to extract
+#	all info needed to 
+#	identify a pile of 
+#	unsorted reads from 
+#	multiple MinION
+#	sequencing 
+#	experiments.
+#
+############
+
+def run(parser, args):
+
+	print "source_filename\ttemplate_fwd_length\tcomplement_rev_length\t2d_length\tasic_id\tasic_temp\theatsink_temp\tchannel\texp_start_time\texp_start_time_string_date\texp_start_time_string_time\tstart_time\tstart_time_string_date\tstart_time_string_time\tduration\tfast5_version"
+
+	for fast5 in Fast5File.Fast5FileSet(args.files):
+
+
+		# run and flowcell parameters
+		asic_temp  = fast5.get_asic_temp()
+		asic_id = fast5.get_asic_id()
+		heatsink_temp = fast5.get_heatsink_temp()
+		channel_number = fast5.get_channel_number()
+
+		# try and get timing info
+		try:
+			start_time = fast5.get_start_time()
+			start_time_string = datetime.datetime.fromtimestamp(float(start_time)).strftime("%Y-%b-%d (%a)\t%H:%M:%S")
+			exp_start_time = fast5.get_exp_start_time()
+			exp_start_time_string = datetime.datetime.fromtimestamp(float(exp_start_time)).strftime("%Y-%b-%d (%a)\t%H:%M:%S")
+			duration = fast5.get_duration()
+		except KeyError:	
+			start_time = "Not found"
+			start_time_string = "NA\tNA"
+			exp_start_time = "Not found"
+			exp_start_time_string = "NA\tNA"
+			duration = "Not found"
+
+		# sequence file info
+		fast5_version = fast5.guess_version()
+
+		# read info
+		fastq_reads = fast5.get_fastqs('all')
+		length_template = None
+		length_complement = None
+		length_2d = None
+		if (len(fastq_reads) > 0):
+			length_template = len(fastq_reads[0].seq)
+		if (len(fastq_reads) > 2):
+			length_complement = len(fastq_reads[1].seq)
+			length_2d = len(fastq_reads[2].seq)
+
+		print "%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s" % (
+			fast5.filename,
+			length_template,
+			length_complement,
+			length_2d,		
+			asic_id, asic_temp, heatsink_temp,channel_number,exp_start_time,exp_start_time_string,start_time,start_time_string,duration,fast5_version)
+
+		fast5.close()

poretools/poretools_main.py

@@ -44,6 +44,8 @@ def run_subtool(parser, args):
         import winner as submodule
     elif args.command == 'yield_plot':
         import yield_plot as submodule
+    elif args.command == 'index':
+        import index as submodule
 
     # run the chosen submodule.
     submodule.run(parser, args)
@@ -289,7 +291,7 @@ def main():
     parser_nucdist.set_defaults(func=run_subtool)
 
     #########
-    # nucdist
+    # metadata
     #########
     parser_metadata = subparsers.add_parser('metadata',
                                         help='Return run metadata such as ASIC ID and temperature from a set of FAST5 files')
@@ -298,6 +300,16 @@ def main():
     parser_metadata.set_defaults(func=run_subtool)
 
 
+    #########
+    # index
+    #########
+    parser_index = subparsers.add_parser('index',
+                                        help='Tabulate all file location info and metadata such as ASIC ID and temperature from a set of FAST5 files')
+    parser_index.add_argument('files', metavar='FILES', nargs='+',
+                             help='The input FAST5 files.')
+    parser_index.set_defaults(func=run_subtool)
+
+
     ##########
     # qualdist
     ##########