add isoform level clustering benchmarking code and source data

Author: eileen-xue

misc/cluster_benchmark.py

@@ -0,0 +1,45 @@
+import csv
+from sklearn import metrics
+import subprocess as sub
+import argparse
+
+parser = argparse.ArgumentParser(description='RATTLE running clustering and cluster summary step')
+parser.add_argument('input', type=str, help='input reads file(required)')
+parser.add_argument('output', type=str, help='output folder (required)')
+parser.add_argument('threads', type=int, help='threads number to run RATTLE (required)')
+parser.add_argument('--rna', action='store_true', help='whether to use RNA mode (instead of cDNA)')
+args = parser.parse_args()
+
+if args.rna:
+    rattle_run = "./rattle cluster -i " + args.input + " -t " + str(args.threads) + " -o " + args.output + " --rna --iso"
+else:
+    rattle_run = "./rattle cluster -i " + args.input + " -t " + str(args.threads) + " -o " + args.output + " --iso"
+p = sub.Popen(rattle_run, shell=True)
+p.wait()
+print("RATTLE isoform-level clustering completed")
+
+rattle_run = "./rattle cluster_summary -i " + args.input + " -c " + args.output + "/clusters.out > " + args.output + "/summary.tsv"
+p = sub.Popen(rattle_run, shell=True)
+p.wait()
+print("RATTLE cluster summary completed")
+
+filename = args.output + "/summary.tsv"
+csv_reader = csv.reader(open(filename))
+tsp = {}
+with open('./toyset/cluster_benchmark/input/sample_ref.fa', 'r') as f:
+    for count, line in enumerate(f, start=1):
+        if count % 2 == 1:
+            l = line.split()
+            tsp[l[0][1:]] = l[3][5:]
+
+labels_true_t = []
+labels_pred =[]
+labels_true_g = []
+for line in csv_reader:
+    labels_true_t.append(line[1])
+    labels_pred.append(line[2])
+    labels_true_g.append(tsp[line[1]])
+print("homogeneity score with (sample reads and human transcriptome) is: {:.2f}%".format(metrics.homogeneity_score(labels_true_t, labels_pred) * 100))
+print("completeness score with (sample reads and human transcriptome) is: {:.2f}%".format(metrics.completeness_score(labels_true_t, labels_pred) * 100))
+print("homogeneity score with (sample reads and human gene) is: {:.2f}%".format(metrics.homogeneity_score(labels_true_g, labels_pred) * 100))
+print("completeness score with (sample reads and human gene) is: {:.2f}%".format(metrics.completeness_score(labels_true_g, labels_pred) * 100))