def format_commands(self): samples = [] if self.by_sample: samples = list(set([basename(x).split("_")[0] for x in self.files])) else: samples = list(set( [basename(x).split(".split")[0] for x in self.files] )) for sample in samples: merge = [] for filename in self.files: if "{}_".format(sample) in filename: merge += [filename] pwd = getcwd() output = self.rebase_file(join(pwd, sample + self.extension)) regex = "\.split\.[a-z]{1,10}?" try: match = search(regex, output).group(0) output = sub(match, "", output) except AttributeError as err: # no attribute 'group' raise (err) combine_files(merge, output)
def format_commands(self): input_f = join(self.input_root, self.all_reads_name) output = join(self.output_root, self.assembly_name) if self.verbose: print("Combining input files...", file=stderr) if not self.dry_run: combine_files(self.files, input_f) else: print("Would have combined files if not dry_run...", file=stderr) job_name = "{}{}".format(self.cluster_options["job_name"], basename(self.all_reads_name)) command = ("tadpole.sh in={} out={} prealloc=t prefilter=2 " "mincontig={} mincov={} k={}").format(input_f, output, self.mincontig, self.mincov, self.kmer_len) self.commands[job_name] = command if self.verbose: print(command, file=stderr)