Mercurial > repos > peterjc > mira_assembler
changeset 3:298f5c1d9521
Uploaded v0.0.2 of the wrapper, which improves capture of the stdout/stderr log file from MIRA.
author | peterjc |
---|---|
date | Tue, 21 Jun 2011 09:50:32 -0400 |
parents | 73263d5c2c9f |
children | 117cce3296af |
files | tools/sr_assembly/mira.py tools/sr_assembly/mira.txt tools/sr_assembly/mira.xml |
diffstat | 3 files changed, 39 insertions(+), 11 deletions(-) [+] |
line wrap: on
line diff
--- a/tools/sr_assembly/mira.py Tue Jun 21 09:48:11 2011 -0400 +++ b/tools/sr_assembly/mira.py Tue Jun 21 09:50:32 2011 -0400 @@ -31,16 +31,21 @@ def collect_output(temp, name): n3 = (temp, name, name, name) - tcs_to_tabular("%s/%s_assembly/%s_d_results/%s_out.tcs" % n3, out_tcs) - for old, new in [("%s/%s_assembly/%s_d_results/%s_out.unpadded.fasta" % n3, out_fasta), - ("%s/%s_assembly/%s_d_results/%s_out.unpadded.fasta.qual" % n3, out_qual), - ("%s/%s_assembly/%s_d_results/%s_out.wig" % n3, out_wig), - ("%s/%s_assembly/%s_d_results/%s_out.caf" % n3, out_caf), - ("%s/%s_assembly/%s_d_results/%s_out.ace" % n3, out_ace)]: + f = "%s/%s_assembly/%s_d_results" % (temp, name, name) + if not os.path.isdir(f): + stop_err("Missing output folder") + if not os.listdir(f): + stop_err("Empty output folder") + for old, new in [("%s/%s_out.unpadded.fasta" % (f, name), out_fasta), + ("%s/%s_out.unpadded.fasta.qual" % (f, name), out_qual), + ("%s/%s_out.wig" % (f, name), out_wig), + ("%s/%s_out.caf" % (f, name), out_caf), + ("%s/%s_out.ace" % (f, name), out_ace)]: if not os.path.isfile(old): stop_err("Missing %s output file" % os.path.splitext(old)[-1]) else: shutil.move(old, new) + tcs_to_tabular("%s/%s_assembly/%s_d_results/%s_out.tcs" % n3, out_tcs) def clean_up(temp, name): folder = "%s/%s_assembly" % (temp, name) @@ -48,22 +53,44 @@ shutil.rmtree(folder) #TODO - Run MIRA in /tmp or a configurable directory? +#Currently Galaxy puts us somewhere safe like: +#/opt/galaxy-dist/database/job_working_directory/846/ temp = "." name, out_fasta, out_qual, out_tcs, out_ace, out_caf, out_wig, out_log = sys.argv[1:9] + start_time = time.time() +cmd = " ".join(sys.argv[9:]) + +assert os.path.isdir(temp) +d = "%s_assembly" % name +assert not os.path.isdir(d) try: - cmd = " ".join(sys.argv[9:]) + #Check path access + os.mkdir(d) +except Exception, err: + sys.stderr.write("Error making directory %s\n%s" % (d, err)) + sys.exit(1) + +#print os.path.abspath(".") +#print cmd + +handle = open(out_log, "w") +try: + #Run MIRA child = subprocess.Popen(sys.argv[9:], - stdout=subprocess.PIPE, stderr=subprocess.STDOUT) + stdout=handle, + stderr=subprocess.STDOUT) except Exception, err: sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err)) + #TODO - call clean up? + handle.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err)) + handle.close() sys.exit(1) #Use .communicate as can get deadlocks with .wait(), stdout, stderr = child.communicate() +assert not stdout and not stderr #Should be empty as sent to handle run_time = time.time() - start_time return_code = child.returncode -handle = open(out_log, "w") -handle.write(stdout) handle.write("\n\nMIRA took %0.2f minutes\n" % (run_time / 60.0)) print "MIRA took %0.2f minutes" % (run_time / 60.0) if return_code:
--- a/tools/sr_assembly/mira.txt Tue Jun 21 09:48:11 2011 -0400 +++ b/tools/sr_assembly/mira.txt Tue Jun 21 09:50:32 2011 -0400 @@ -34,6 +34,7 @@ ======= v0.0.1 - Initial version (working prototype) +v0.0.2 - Improve capture of stdout/stderr (should see it as it runs) Developers
--- a/tools/sr_assembly/mira.xml Tue Jun 21 09:48:11 2011 -0400 +++ b/tools/sr_assembly/mira.xml Tue Jun 21 09:50:32 2011 -0400 @@ -1,4 +1,4 @@ -<tool id="mira_assembler" name="Assemble with MIRA" version="0.0.1"> +<tool id="mira_assembler" name="Assemble with MIRA" version="0.0.2"> <description>Takes Sanger, Roche, and Illumina data</description> <command interpreter="python">mira.py mira $out_fasta $out_qual $out_tcs $out_ace $out_caf $out_wig $out_log ##Give the wrapper script list of output filenames, then the mira command...