add argparse for combine_2.py

README.md

@@ -140,21 +140,19 @@ In this example, we request one job per combination of (i, j, k). Hence there wi
 Once all the jobs are complete, combine the results together to generate a final score list.
 Note this step does not need to run using SLURM scheduler.
 ```bash
-python3 ~/.local/lib/python3.6/site-packages/silhouetteRank/combine.py -i . -o silhouette_overall_pval.txt -l master.combine.log -v -r 0.95 0.99 -e 0.005 0.01 0.05 0.1 0.3 -m dissim 
+python3 ~/.local/lib/python3.6/site-packages/silhouetteRank/combine_2.py -i . -v -r 0.95 0.99 -e 0.005 0.01 0.05 0.1 0.3 -m dissim 
 ```
 
 **Table 2**. Explanations of `combine.py`: 
 | param | explanation |
 | ----- | ------------------ |
 | -i | The results directory (containing directories like result_5000_0.95_0.300) |
-| -o | Output file |
-| -l | Log file name (just file name, no path) |
 | -v | Verbose |
 | -r | Array. Float. Local spatial distance weighting constant. (recommend 0.95 - 0.995) |
 | -e | Array. Float. Top proportion of cells to binarize to 1 (0 - 1) |
 | -m | dissim or sim. Use dissimilarity matrix (default) or similarity matrix |
 
-The final result is saved in the file `silhouette.overall.pval.txt`.
+The final result is saved in the file `silhouette.overall.pval.txt` located in the results directory.
 
 Note: the parameter settings for `-r` and `-e` work well for most datasets. Users can safely use the same commands for all datasets.
 

silhouetteRank/combine_2.py

@@ -0,0 +1,70 @@
+import math
+import sys
+import os
+import re
+import scipy
+import scipy.stats
+import numpy as np
+from operator import itemgetter
+import silhouetteRank
+import silhouetteRank.prep as prep
+import silhouetteRank.evaluate_exact_one_2b as evaluate_exact_one_2b
+import silhouetteRank.use_previous_cluster as use_previous_cluster
+import silhouetteRank.combine as combine
+import logging
+import argparse
+import subprocess
+
+def read(n):
+	f = open(n)
+	by_gene = {}
+	for l in f:
+		l = l.rstrip("\n")
+		ll = l.split()
+		gene = ll[0]
+		pval = float(ll[-2])
+		by_gene[gene] = pval
+	f.close()
+	return by_gene
+
+def do_one(args):
+	result = subprocess.call("Rscript --version 2> /dev/null", shell=True)
+	if result==127:
+		sys.stderr.write("Rscript is not found\n")
+		sys.stderr.flush()
+		sys.exit(1)	
+
+	check_required = ["expr.npy", "Xcen.npy", "genes.npy"]
+	for cr in check_required:
+		if not os.path.isfile("%s/%s" % (args.input, cr)):
+			sys.stderr.write("%s file does not exist\n" % cr)
+			sys.stderr.flush()
+			sys.exit(1)
+
+	for rbp_p in args.rbp_ps:
+		for examine_top in args.examine_tops:
+			random_dir = "%s/result_sim_5000_%.2f_%.3f" % (args.input, rbp_p, examine_top)
+			score_file = "%s/silhouette.sim.exact.rbp.%.2f.top.%.3f.txt" % (args.input, rbp_p, examine_top)
+			output_score_file = "%s/silhouette.sim.exact.rbp.%.2f.top.%.3f.pval.txt" % (args.input, rbp_p, examine_top)
+			if args.matrix_type=="dissim":
+				random_dir = "%s/result_5000_%.2f_%.3f" % (args.input, rbp_p, examine_top)
+				score_file = "%s/silhouette.exact.rbp.%.2f.top.%.3f.txt" % (args.input, rbp_p, examine_top)
+				output_score_file = "%s/silhouette.exact.rbp.%.2f.top.%.3f.pval.txt" % (args.input, rbp_p, examine_top)
+			args1 = argparse.Namespace(expr="../expression.txt", centroid="../Xcen.good", examine_top=examine_top, input=score_file, input_random=random_dir, output=output_score_file, outdir=args.input, query_sizes=args.query_sizes, overwrite_input_bin=False, verbose=verbose, log_file="master.pvalue.log")
+			use_previous_cluster.do_one(args1)
+
+	combined_file = "%s/silhouette.overall.pval.txt" % args.input
+	if args.matrix_type=="sim":
+		combined_file = "%s/silhouette.sim.overall.pval.txt" % args.input
+	args1 = argparse.Namespace(rbp_ps=args.rbp_ps, examine_tops=args.examine_tops, matrix_type=args.matrix_type, input=args.input, output=combined_file)
+	combine.do_one(args1)
+
+if __name__=="__main__":
+	parser = argparse.ArgumentParser(description="combine.py: combine spatial scores across parameters", formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+	parser.add_argument("-r", "--rbp-ps", dest="rbp_ps", nargs="+", type=float, default=[0.95, 0.99], help="p parameter of RBP")
+	parser.add_argument("-e", "--examine-tops", dest="examine_tops", nargs="+", type=float, default=[0.005, 0.010, 0.050, 0.100, 0.300], help="top proportion of cells per gene to be 1's (expressed)")
+	parser.add_argument("-m", "--matrix-type", dest="matrix_type", type=str, choices=["sim", "dissim"], help="whether to calculate similarity matrix or dissimilarity matrix", default="dissim")
+	parser.add_argument("-i", "--input-dir", dest="input", type=str, default=".", help="input directory containing individual spatial score rankings (to be aggregated)")
+	#parser.add_argument("-o", "--output", dest="output", type=str, required=True, help="output file name")
+	args = parser.parse_args()
+	do_one(args)