Mercurial > repos > rnateam > graphclust_postprocessing
annotate evaluation.py @ 1:e166d1382033 draft
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 11e50007837b1efa01a3039c92df0ebf63f0f7e9
| author | rnateam |
|---|---|
| date | Thu, 22 Dec 2016 08:49:14 -0500 |
| parents | 4503c49f31c4 |
| children | 6c88ad83de28 |
| 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 def sh(script): |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
6 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
|
7 |
|
1
e166d1382033
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 11e50007837b1efa01a3039c92df0ebf63f0f7e9
rnateam
parents:
0
diff
changeset
|
8 dataNames = "FASTA/data.names" |
|
0
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 listOfClusters = [] |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
11 listOfClasses = [] |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
12 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
|
13 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
|
14 |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
15 blackList = [] |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
16 numberOfClusters = 0 |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
17 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
|
18 numberOfClusters += 1 |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
19 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
|
20 for line in f.readlines(): |
|
1
e166d1382033
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 11e50007837b1efa01a3039c92df0ebf63f0f7e9
rnateam
parents:
0
diff
changeset
|
21 uniqueId = line.split()[7] |
|
0
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
22 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
|
23 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
|
24 listOfClasses.append(rnaClass) |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
25 listOfClusters.append(clustNum) |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
26 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
|
27 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
|
28 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
|
29 rnaClass, sep, tail = fullUniqeId.partition("_") |
|
1
e166d1382033
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 11e50007837b1efa01a3039c92df0ebf63f0f7e9
rnateam
parents:
0
diff
changeset
|
30 if fullUniqeId == uniqueId: |
|
0
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
31 blackList.append(uniqueId) |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
32 |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
33 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
|
34 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
|
35 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
|
36 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
|
37 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
|
38 rnaClass, sep, tail = fullUniqeId.partition("_") |
|
1
e166d1382033
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 11e50007837b1efa01a3039c92df0ebf63f0f7e9
rnateam
parents:
0
diff
changeset
|
39 if fullUniqeId not in blackList: |
|
0
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
40 listOfClasses.append(rnaClass) |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
41 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
|
42 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
|
43 |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
44 toWrite = "" |
|
4503c49f31c4
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 21aaee40723b5341b4236edeb0e72995c2054053
rnateam
parents:
diff
changeset
|
45 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
|
46 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
|
47 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
|
48 full.write(toWrite) |