def run_clustering(infile, outfile):
'''Run clustering algorithms on tSNE dim reduced data'''
job_memory = "20G"
pipeline_dir = PARAMS["pipeline_dir"]
clust_methods = PARAMS["analysis_clust_methods"].split(",")
# Rphenograph, ClusterX, FlowSOM
statements = []
for clust in clust_methods:
outname = outfile.replace(".cluster.touch",
".") + str(clust) + ".RData"
log = outfile.replace(".cluster.touch", ".") + str(clust) + ".log"
if clust == "Rphenograph":
infile = infile.replace(
".tsne.RData",
".norm.RData") # runs on high dimensionality data
no_clust = " "
print("Rphenograph")
elif clust == "FlowSOM":
infile = infile.replace(
".tsne.RData",
".norm.RData") # runs on high dimensionality data
no_clust = "--noClusters " + str(PARAMS["analysis_clust_no"])
print("FlowSOM")
elif clust == "ClusterX":
infile = infile # run on tSNE reduced data
no_clust = " "
print("ClusterX")
else:
print("Specify clustering method in pipeline.ini")
statements.append('''Rscript %(pipeline_dir)s/R/CYTOFclust.R
--infile %(infile)s
--outfile %(outname)s
--clusterMethod %(clust)s
%(no_clust)s
&> %(log)s''' % locals())
print(statements)
P.run()
IOTools.touchFile(outfile)
def run_tsne(infiles, outfile):
'''Process FCS data and run tSNE. Supports multiple options for
tsne parameters "perplexity" and "iterations" '''
job_memory = "40G"
pipeline_dir = PARAMS["pipeline_dir"]
infiles = ','.join(infiles)
marker_list = PARAMS[
"analysis_markers"] # markers to use for tsne & downstream analysis
perplexity = str(PARAMS["tsne_perplexity"]).split(",")
iterations = str(PARAMS["tsne_iterations"]).split(",")
events = PARAMS["analysis_no_events"] # no events per sample
statements = []
for p in perplexity:
for i in iterations:
out_norm = outfile.replace(
".matrix.touch",
"") + "." + str(p) + "_" + str(i) + ".norm.RData"
out_tsne = outfile.replace(
".matrix.touch",
"") + "." + str(p) + "_" + str(i) + ".tsne.RData"
log = outfile.replace(".touch", ".log")
statements.append('''Rscript %(pipeline_dir)s/R/cytofkit.R
--infiles %(infiles)s
--out_norm %(out_norm)s
--out_tsne %(out_tsne)s
--markers %(marker_list)s
--perplexity %(p)s
--iterations %(i)s
--no_events %(events)s
&> %(log)s''' % locals())
print(statements)
P.run()
IOTools.touchFile(outfile) # creates sentinel file for task monitoring
def compute_file_metrics(infile, outfile, metric, suffixes):
"""apply a tool to compute metrics on a list of files matching
regex_pattern."""
if suffixes is None or len(suffixes) == 0:
E.info("No metrics computed for {}".format(outfile))
IOTools.touchFile(outfile)
return
track = P.snip(infile, ".log")
# convert regex patterns to a suffix match:
# prepend a .*
# append a $
regex_pattern = " -or ".join(["-regex .*{}$".format(pipes.quote(x))
for x in suffixes])
E.debug("applying metric {} to files matching {}".format(metric,
regex_pattern))
if metric == "file":
statement = '''find %(track)s.dir
-type f
-not -regex '.*\/report.*'
-not -regex '.*\/_.*'
\( %(regex_pattern)s \)
| sort -k1,1
> %(outfile)s'''
else:
statement = '''find %(track)s.dir
-type f
-not -regex '.*\/report.*'
-not -regex '.*\/_.*'
\( %(regex_pattern)s \)
-exec %(pipeline_scriptsdir)s/cgat_file_apply.sh {} %(metric)s \;
| perl -p -e "s/ +/\\t/g"
| sort -k1,1
> %(outfile)s'''
P.run()
