--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/multigps.py	Sat Jan 02 14:07:37 2016 -0500
@@ -0,0 +1,193 @@
+import argparse
+import os
+import shutil
+import subprocess
+import sys
+import tempfile
+
+BUFF_SIZE = 1048576
+DESIGN_FILE = 'design.tabular'
+
+
+def generate_design_file(input_items, filename):
+    design_file = open(filename, 'wb')
+    for items in input_items:
+        filename, ext, signal_control, condition_name, replicate_name, read_distribution_file, fixed_read = items
+        # MultiGPS version 0.5 does not support the newer scidx datatype.
+        if ext == 'scidx':
+            datatype = 'IDX'
+        else:
+            datatype = ext.upper()
+        design_file.write('%s\n' % '\t'.join([filename, signal_control, datatype, condition_name, replicate_name, read_distribution_file, fixed_read]))
+    design_file.close()
+
+
+def get_stderr_exception(tmp_err, tmp_stderr, tmp_out, tmp_stdout, include_stdout=False):
+    tmp_stderr.close()
+    # Get stderr, allowing for case where it's very large.
+    tmp_stderr = open(tmp_err, 'rb')
+    stderr_str = ''
+    buffsize = BUFF_SIZE
+    try:
+        while True:
+            stderr_str += tmp_stderr.read(buffsize)
+            if not stderr_str or len(stderr_str) % buffsize != 0:
+                break
+    except OverflowError:
+        pass
+    tmp_stderr.close()
+    if include_stdout:
+        tmp_stdout = open(tmp_out, 'rb')
+        stdout_str = ''
+        buffsize = BUFF_SIZE
+        try:
+            while True:
+                stdout_str += tmp_stdout.read(buffsize)
+                if not stdout_str or len(stdout_str) % buffsize != 0:
+                    break
+        except OverflowError:
+            pass
+    tmp_stdout.close()
+    if include_stdout:
+        return 'STDOUT\n%s\n\nSTDERR\n%s\n' % (stdout_str, stderr_str)
+    return stderr_str
+
+
+def stop_err(msg):
+    sys.stderr.write(msg)
+    sys.exit(1)
+
+parser = argparse.ArgumentParser()
+parser.add_argument('--input_item', dest='input_items', action='append', nargs=7, help="Input datasets and associated parameters")
+parser.add_argument('--threads', dest='threads', type=int, default=4, help='The number of threads to run')
+parser.add_argument('--multigps_jar', dest='multigps_jar', help='Path to the MultiGPS jar file')
+parser.add_argument('--geninfo', dest='geninfo', help='File listing the lengths of all chromosomes')
+parser.add_argument('--use_motif', dest='use_motif', help='Perform motif-finding or use a motif-prior')
+parser.add_argument('--seq_dir', dest='seq_dir', default=None, help='Directory containing fasta files corresponding to every named chromosome')
+parser.add_argument('--positional_prior', dest='positional_prior', default=None, help='Perform inter-experiment positional prior')
+parser.add_argument('--events_shared_probability', dest='events_shared_probability', type=float, default=0.9, help='Probability that events are shared across conditions')
+parser.add_argument('--motifs', dest='motifs', default=None, help='Perform motif-finding and motif priors')
+parser.add_argument('--motif_finding_only', dest='motif_finding_only', default=None, help='Perform motif-finding only')
+parser.add_argument('--num_motifs', dest='num_motifs', type=int, default=None, help='Number of motifs MEME should find for each condition')
+parser.add_argument('--mememinw', dest='mememinw', type=int, default=None, help='Minimum motif width for MEME')
+parser.add_argument('--mememaxw', dest='mememaxw', type=int, default=None, help='Maximum motif width for MEME')
+parser.add_argument('--fixedpb', dest='fixedpb', type=int, default=None, help='Fixed per-base limit')
+parser.add_argument('--poissongausspb', dest='poissongausspb', type=int, default=None, help='Poisson threshold for filtering per base')
+parser.add_argument('--non_unique_reads', dest='non_unique_reads', default=None, help='Use non-unique reads')
+parser.add_argument('--minqvalue', dest='minqvalue', type=int, default=None, help='Minimum Q-value (corrected p-value) of reported binding events')
+parser.add_argument('--minfold', dest='minfold', type=int, default=None, help='Minimum event fold-change vs scaled control')
+parser.add_argument('--diff_enrichment_tests', dest='diff_enrichment_tests', help='Run differential enrichment tests')
+parser.add_argument('--edgerod', dest='edgerod', type=int, default=None, help='EdgeR over-dispersion parameter value')
+parser.add_argument('--diffp', dest='diffp', type=int, default=None, help='Minimum p-value for reporting differential enrichment')
+parser.add_argument('--noscaling', dest='noscaling', default=None, help='Do not use signal vs control scaling')
+parser.add_argument('--medianscale', dest='medianscale', default=None, help='Use the median signal/control ratio as the scaling factor')
+parser.add_argument('--sesscale', dest='sesscale', default=None, help='Estimate scaling factor by SES')
+parser.add_argument('--scalewin', dest='scalewin', type=int, default=None, help='Window size for estimating scaling ratios')
+parser.add_argument('--max_training_rounds', dest='max_training_rounds', type=int, default=None, help='Maximum number of training rounds for updating binding event read distributions')
+parser.add_argument('--exclude_file', dest='exclude_file', default=None, help='File containing a set of regions to ignore during MultiGPS training')
+parser.add_argument('--binding_model_updates', dest='binding_model_updates', default=None, help='Perform binding model updates')
+parser.add_argument('--minmodelupdateevents', dest='minmodelupdateevents', type=int, default=None, help='Minimum number of events to support an update of the read distribution')
+parser.add_argument('--binding_model_smoothing', dest='binding_model_smoothing', default=None, help='Binding model smoothing')
+parser.add_argument('--spline_smooth', dest='spline_smooth', type=int, default=None, help='Binding model smoothing value')
+parser.add_argument('--gauss_smooth', dest='gauss_smooth', type=int, default=None, help='Gaussian smoothing std dev')
+parser.add_argument('--joint_in_model', dest='joint_in_model', default=None, help='Allow joint events in model updates')
+parser.add_argument('--ml_config_not_shared', dest='ml_config_not_shared', default=None, help='Share component configs in the ML step')
+parser.add_argument('--output_html_path', dest='output_html_path', help='Output html results file')
+parser.add_argument('--output_html_files_path', dest='output_html_files_path', help='Output html extra files')
+parser.add_argument('--output_process_path', dest='output_process_path', default=None, help='Output file for capturing stdout')
+args = parser.parse_args()
+
+try:
+    # Preparation.
+    tmp_dir = tempfile.mkdtemp(prefix="tmp-multigps-")
+    generate_design_file(args.input_items, DESIGN_FILE)
+    # Build the command line.
+    cmd_list = ['java -jar %s --verbose' % args.multigps_jar]
+    cmd_list.append('--threads %d' % args.threads)
+    cmd_list.append('--geninfo %s' % args.geninfo)
+    if args.use_motif == 'yes':
+        cmd_list.append('--seq %s' % args.seq_dir)
+        if args.positional_prior == 'no':
+            cmd_list.append('--noposprior')
+        cmd_list.append('--probshared %3f' % args.events_shared_probability)
+        if args.motifs == 'no':
+            cmd_list.append('--nomotifs')
+        if args.motif_finding_only == 'yes':
+            cmd_list.append('--nomotifprior')
+        if args.num_motifs is not None:
+            cmd_list.append('--memenmotifs %d' % args.num_motifs)
+        if args.mememinw is not None:
+            cmd_list.append('--mememinw %d' % args.mememinw)
+        if args.mememaxw is not None:
+            cmd_list.append('--mememaxw %d' % args.mememaxw)
+    if args.fixedpb is not None:
+        cmd_list.append('--fixedpb %d' % args.fixedpb)
+    if args.poissongausspb is not None:
+        cmd_list.append('--poissongausspb %d' % args.poissongausspb)
+    if args.non_unique_reads == 'yes':
+        cmd_list.append('--nonunique')
+    if args.minqvalue is not None:
+        cmd_list.append('--q %d' % args.minqvalue)
+    if args.minfold is not None:
+        cmd_list.append('--minfold %d' % args.minfold)
+    if args.diff_enrichment_tests == 'no':
+        cmd_list.append('--nodifftests')
+    if args.edgerod is not None:
+        cmd_list.append('--edgerod %d' % args.edgerod)
+    if args.diffp is not None:
+        cmd_list.append('--diffp %d' % args.diffp)
+    if args.noscaling == 'no':
+        cmd_list.append('--noscaling')
+    if args.medianscale == "yes":
+        cmd_list.append('--medianscale')
+    if args.sesscale == "yes":
+        cmd_list.append('--sesscale')
+    if args.scalewin is not None:
+        cmd_list.append('--scalewin %d' % args.scalewin)
+    if args.max_training_rounds is not None:
+        cmd_list.append('--r %d' % args.max_training_rounds)
+    if args.exclude_file not in [None, 'None']:
+        cmd_list.append('--exclude %s' % args.exclude_file)
+    if args.binding_model_updates == 'no':
+        cmd_list.append('--nomodelupdate')
+    if args.minmodelupdateevents is not None:
+        cmd_list.append('--minmodelupdateevents %d' % args.minmodelupdateevents)
+    if args.binding_model_smoothing == 'yes':
+        if args.spline_smooth is not None:
+            cmd_list.append('--splinesmoothparam %d' % args.spline_smooth)
+    else:
+        cmd_list.append('--nomodelsmoothing')
+        if args.gauss_smooth is not None:
+            cmd_list.append('--gaussmodelsmoothing')
+            cmd_list.append('--gausssmoothparam %d' % args.gauss_smooth)
+    if args.joint_in_model == 'yes':
+        cmd_list.append('--jointinmodel')
+    if args.ml_config_not_shared == 'no':
+        cmd_list.append('--mlconfignotshared')
+    cmd_list.append('--design %s' % DESIGN_FILE)
+    cmd_list.append('--out %s' % args.output_html_files_path)
+    if args.output_process_path is None:
+        cmd_list.append('2>/dev/null')
+    else:
+        cmd_list.append('2>%s' % args.output_process_path)
+    cmd = ' '.join(cmd_list)
+    # Define command response buffers.
+    tmp_out = tempfile.NamedTemporaryFile(dir=tmp_dir).name
+    tmp_stdout = open(tmp_out, 'wb')
+    tmp_err = tempfile.NamedTemporaryFile(dir=tmp_dir).name
+    tmp_stderr = open(tmp_err, 'wb')
+    tmp_filename = tempfile.NamedTemporaryFile(dir=tmp_dir).name
+    # Execute the command.
+    proc = subprocess.Popen(args=cmd, stderr=tmp_stderr, stdout=tmp_stdout, shell=True)
+    rc = proc.wait()
+    if rc != 0:
+        error_message = get_stderr_exception(tmp_err, tmp_stderr, tmp_out, tmp_stdout)
+        stop_err(error_message)
+    # Move the output HTML file to the output dataset path.
+    output_html_filename = 'multiGPS_%s_results.html' % os.path.split(args.output_html_files_path)[1]
+    shutil.move(os.path.join(args.output_html_files_path, output_html_filename), args.output_html_path)
+except Exception, e:
+    stop_err('Error running MultiGPS\n%s\n' % str(e))
+finally:
+    if os.path.exists(tmp_dir):
+        shutil.rmtree(tmp_dir)