Mercurial > repos > rnateam > graphclust_postprocessing
annotate evaluation.py @ 0:4503c49f31c4 draft
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
author | rnateam |
---|---|
date | Fri, 16 Dec 2016 07:32:23 -0500 |
parents | |
children | e166d1382033 |
rev | line source |
---|---|
0
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
1 import glob |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
2 from os import system |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
3 import re |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
4 |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
5 |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
6 def sh(script): |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
7 system("bash -c '%s'" % script) |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
8 |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
9 |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
10 dataNames = "FASTA/data.names" |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
11 listOfClusters = [] |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
12 listOfClasses = [] |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
13 cluster_seqs_stats_path = "RESULTS/*.cluster.all" |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
14 cluster_seqs_stats_files = glob.glob(cluster_seqs_stats_path) |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
15 |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
16 blackList = [] |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
17 numberOfClusters = 0 |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
18 for singleFile in sorted(cluster_seqs_stats_files): |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
19 numberOfClusters += 1 |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
20 with open(singleFile, "r") as f: |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
21 for line in f.readlines(): |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
22 uniqueId = line.split()[6] |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
23 clustNum = line.split()[1] |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
24 rnaClass, sep, tail = uniqueId.partition("_") |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
25 listOfClasses.append(rnaClass) |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
26 listOfClusters.append(clustNum) |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
27 with open(dataNames, "r") as names: |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
28 for line in names.readlines(): |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
29 fullUniqeId = line.split()[3] |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
30 rnaClass, sep, tail = fullUniqeId.partition("_") |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
31 short_unique = re.findall("_".join(["[^_]+"] * 2), fullUniqeId)[0] |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
32 if short_unique == uniqueId: |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
33 blackList.append(uniqueId) |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
34 |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
35 numberOfClusters += 1 # 1 cluster for all unassigned seqs |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
36 with open(dataNames, "r") as names: |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
37 for line in names.readlines(): |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
38 fullUniqeId = line.split()[3] |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
39 rnaClass, sep, tail = fullUniqeId.partition("_") |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
40 short_unique = re.findall("_".join(["[^_]+"] * 2), fullUniqeId)[0] |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
41 rnaClass, sep, tail = fullUniqeId.partition("_") |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
42 if short_unique not in blackList: |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
43 listOfClasses.append(rnaClass) |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
44 listOfClusters.append(str(numberOfClusters)) |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
45 numberOfClusters += 1 # separate cluster for all unassigned seqs |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
46 |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
47 toWrite = "" |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
48 for i in range(len(listOfClusters)): |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
49 toWrite += listOfClasses[i] + "\t" + listOfClusters[i] + '\n' |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
50 with open("RESULTS/fullTab.tabular", "w") as full: |
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
51 full.write(toWrite) |