Mercurial > repos > rnateam > graphclust_postprocessing

changeset 1:ed8c7191b322 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 11e50007837b1efa01a3039c92df0ebf63f0f7e9
author rnateam
date Thu, 22 Dec 2016 09:06:48 -0500
parents b797e13169a0
children b8e32e577597
files evaluation.py test-data/RESULTS/1.cluster.all test-data/RESULTS/2.cluster.all test-data/RESULTS/partitions/final_partition.used_cmsearch
diffstat 4 files changed, 51 insertions(+), 54 deletions(-) [+]
line wrap: on
line diff
--- a/evaluation.py	Fri Dec 16 07:34:49 2016 -0500
+++ b/evaluation.py	Thu Dec 22 09:06:48 2016 -0500
@@ -2,12 +2,11 @@
 from os import system
 import re
 
-
 def sh(script):
     system("bash -c '%s'" % script)
 
+dataNames = "FASTA/data.names"
 
-dataNames = "FASTA/data.names"
 listOfClusters = []
 listOfClasses = []
 cluster_seqs_stats_path = "RESULTS/*.cluster.all"
@@ -19,7 +18,7 @@
     numberOfClusters += 1
     with open(singleFile, "r") as f:
         for line in f.readlines():
-            uniqueId = line.split()[6]
+            uniqueId = line.split()[7]
             clustNum = line.split()[1]
             rnaClass, sep, tail = uniqueId.partition("_")
             listOfClasses.append(rnaClass)
@@ -28,8 +27,7 @@
                 for line in names.readlines():
                     fullUniqeId = line.split()[3]
                     rnaClass, sep, tail = fullUniqeId.partition("_")
-                    short_unique = re.findall("_".join(["[^_]+"] * 2), fullUniqeId)[0]
-                    if short_unique == uniqueId:
+                    if fullUniqeId == uniqueId:
                         blackList.append(uniqueId)
 
 numberOfClusters += 1  # 1 cluster for all unassigned seqs
@@ -37,9 +35,8 @@
     for line in names.readlines():
         fullUniqeId = line.split()[3]
         rnaClass, sep, tail = fullUniqeId.partition("_")
-        short_unique = re.findall("_".join(["[^_]+"] * 2), fullUniqeId)[0]
         rnaClass, sep, tail = fullUniqeId.partition("_")
-        if short_unique not in blackList:
+        if fullUniqeId not in blackList:
             listOfClasses.append(rnaClass)
             listOfClusters.append(str(numberOfClusters))
             numberOfClusters += 1  # separate cluster for all unassigned seqs
--- a/test-data/RESULTS/1.cluster.all	Fri Dec 16 07:34:49 2016 -0500
+++ b/test-data/RESULTS/1.cluster.all	Thu Dec 22 09:06:48 2016 -0500
@@ -1,33 +1,33 @@
-CLUSTER  1  CM_SCORE  84.9  MODEL     1.1  RF00005_rep.34  
-CLUSTER  1  CM_SCORE  84.9  MODEL     1.1  RF00005_rep.3   
-CLUSTER  1  CM_SCORE  83.8  MODEL     1.1  RF00005_rep.14  
-CLUSTER  1  CM_SCORE  82.5  MODEL     1.1  RF00005_rep.35  
-CLUSTER  1  CM_SCORE  82.4  MODEL     1.1  RF00005_rep.31  
-CLUSTER  1  CM_SCORE  81.0  MODEL     1.1  RF00005_rep.12  
-CLUSTER  1  CM_SCORE  80.9  MODEL     1.1  RF00005_rep.5   
-CLUSTER  1  CM_SCORE  78.9  MODEL     1.1  RF00005_rep.21  
-CLUSTER  1  CM_SCORE  76.6  MODEL     1.1  RF00005_rep.1   
-CLUSTER  1  CM_SCORE  76.3  MODEL     1.1  RF00005_rep.27  
-CLUSTER  1  CM_SCORE  74.7  MODEL     1.1  RF00005_rep.20  
-CLUSTER  1  CM_SCORE  73.6  MODEL     1.1  RF00005_rep.22  
-CLUSTER  1  CM_SCORE  73.3  MODEL     1.1  RF00005_rep.15  
-CLUSTER  1  CM_SCORE  72.3  MODEL     1.1  RF00005_rep.33  
-CLUSTER  1  CM_SCORE  71.7  MODEL     1.1  RF00005_rep.38  
-CLUSTER  1  CM_SCORE  69.3  MODEL     1.1  RF00005_rep.36  
-CLUSTER  1  CM_SCORE  21.4  MODEL     1.1  RF00005_rep.41  
-CLUSTER  1  CM_SCORE  88.8  CMSEARCH  1.1  RF00005_rep.16  
-CLUSTER  1  CM_SCORE  87.5  CMSEARCH  1.1  RF00005_rep.18  
-CLUSTER  1  CM_SCORE  84.8  CMSEARCH  1.1  RF00005_rep.17  
-CLUSTER  1  CM_SCORE  80.0  CMSEARCH  1.1  RF00005_rep.2   
-CLUSTER  1  CM_SCORE  79.6  CMSEARCH  1.1  RF00005_rep.10  
-CLUSTER  1  CM_SCORE  79.0  CMSEARCH  1.1  RF00005_rep.28  
-CLUSTER  1  CM_SCORE  78.5  CMSEARCH  1.1  RF00005_rep.39  
-CLUSTER  1  CM_SCORE  78.4  CMSEARCH  1.1  RF00005_rep.9   
-CLUSTER  1  CM_SCORE  76.3  CMSEARCH  1.1  RF00005_rep.25  
-CLUSTER  1  CM_SCORE  75.1  CMSEARCH  1.1  RF00005_rep.0   
-CLUSTER  1  CM_SCORE  73.3  CMSEARCH  1.1  RF00005_rep.24  
-CLUSTER  1  CM_SCORE  69.4  CMSEARCH  1.1  RF00005_rep.23  
-CLUSTER  1  CM_SCORE  69.3  CMSEARCH  1.1  RF00005_rep.4   
-CLUSTER  1  CM_SCORE  47.8  CMSEARCH  1.1  RF00005_rep.26  
-CLUSTER  1  CM_SCORE  45.6  CMSEARCH  1.1  RF00005_rep.19  
-CLUSTER  1  CM_SCORE  34.4  CMSEARCH  1.1  RF00005_rep.32  
+CLUSTER  1  CM_SCORE  84.9  MODEL     1.1  ORIGID  RF00005_rep.3_Z54587.1/126-45_37             ORIGHEAD  RF00005_rep.3   
+CLUSTER  1  CM_SCORE  84.9  MODEL     1.1  ORIGID  RF00005_rep.34_AC008443.10/43006-42934_31    ORIGHEAD  RF00005_rep.34  
+CLUSTER  1  CM_SCORE  83.8  MODEL     1.1  ORIGID  RF00005_rep.14_AL021808.2/65570-65498_9      ORIGHEAD  RF00005_rep.14  
+CLUSTER  1  CM_SCORE  82.5  MODEL     1.1  ORIGID  RF00005_rep.35_AC005783.1/27398-27326_32     ORIGHEAD  RF00005_rep.35  
+CLUSTER  1  CM_SCORE  82.4  MODEL     1.1  ORIGID  RF00005_rep.31_AC092686.3/29631-29561_28     ORIGHEAD  RF00005_rep.31  
+CLUSTER  1  CM_SCORE  81.0  MODEL     1.1  ORIGID  RF00005_rep.12_AC108081.2/59868-59786_7      ORIGHEAD  RF00005_rep.12  
+CLUSTER  1  CM_SCORE  80.9  MODEL     1.1  ORIGID  RF00005_rep.5_AL590385.23/26129-26058_46     ORIGHEAD  RF00005_rep.5   
+CLUSTER  1  CM_SCORE  78.9  MODEL     1.1  ORIGID  RF00005_rep.21_AL355149.13/15278-15208_17    ORIGHEAD  RF00005_rep.21  
+CLUSTER  1  CM_SCORE  76.6  MODEL     1.1  ORIGID  RF00005_rep.1_AC005329.1/7043-6971_15        ORIGHEAD  RF00005_rep.1   
+CLUSTER  1  CM_SCORE  76.3  MODEL     1.1  ORIGID  RF00005_rep.27_AL352978.6/119697-119770_23   ORIGHEAD  RF00005_rep.27  
+CLUSTER  1  CM_SCORE  74.7  MODEL     1.1  ORIGID  RF00005_rep.20_AL671879.2/100356-100285_16   ORIGHEAD  RF00005_rep.20  
+CLUSTER  1  CM_SCORE  73.6  MODEL     1.1  ORIGID  RF00005_rep.22_AL590385.23/26487-26416_18    ORIGHEAD  RF00005_rep.22  
+CLUSTER  1  CM_SCORE  73.3  MODEL     1.1  ORIGID  RF00005_rep.15_AC008443.10/42590-42518_10    ORIGHEAD  RF00005_rep.15  
+CLUSTER  1  CM_SCORE  72.3  MODEL     1.1  ORIGID  RF00005_rep.33_AC018638.5/4694-4623_30       ORIGHEAD  RF00005_rep.33  
+CLUSTER  1  CM_SCORE  71.7  MODEL     1.1  ORIGID  RF00005_rep.38_J00309.1/356-427_35           ORIGHEAD  RF00005_rep.38  
+CLUSTER  1  CM_SCORE  69.3  MODEL     1.1  ORIGID  RF00005_rep.36_AC007298.17/145366-145295_33  ORIGHEAD  RF00005_rep.36  
+CLUSTER  1  CM_SCORE  21.4  MODEL     1.1  ORIGID  RF00005_rep.41_AC093311.2/140036-139968_39   ORIGHEAD  RF00005_rep.41  
+CLUSTER  1  CM_SCORE  88.8  CMSEARCH  1.1  ORIGID  RF00005_rep.16_AL133551.13/12355-12436_11    ORIGHEAD  RF00005_rep.16  
+CLUSTER  1  CM_SCORE  87.5  CMSEARCH  1.1  ORIGID  RF00005_rep.18_AL021918.1/81116-81197_13     ORIGHEAD  RF00005_rep.18  
+CLUSTER  1  CM_SCORE  84.8  CMSEARCH  1.1  ORIGID  RF00005_rep.17_AL021918.1/54817-54736_12     ORIGHEAD  RF00005_rep.17  
+CLUSTER  1  CM_SCORE  80.0  CMSEARCH  1.1  ORIGID  RF00005_rep.2_AL662865.4/12206-12135_26      ORIGHEAD  RF00005_rep.2   
+CLUSTER  1  CM_SCORE  79.6  CMSEARCH  1.1  ORIGID  RF00005_rep.10_X58792.1/174-245_5            ORIGHEAD  RF00005_rep.10  
+CLUSTER  1  CM_SCORE  79.0  CMSEARCH  1.1  ORIGID  RF00005_rep.28_X04779.1/1-73_24              ORIGHEAD  RF00005_rep.28  
+CLUSTER  1  CM_SCORE  78.5  CMSEARCH  1.1  ORIGID  RF00005_rep.39_AL031229.2/40502-40430_36     ORIGHEAD  RF00005_rep.39  
+CLUSTER  1  CM_SCORE  78.4  CMSEARCH  1.1  ORIGID  RF00005_rep.9_AP000442.6/2022-1950_50        ORIGHEAD  RF00005_rep.9   
+CLUSTER  1  CM_SCORE  76.3  CMSEARCH  1.1  ORIGID  RF00005_rep.25_AC006449.19/196857-196784_21  ORIGHEAD  RF00005_rep.25  
+CLUSTER  1  CM_SCORE  75.1  CMSEARCH  1.1  ORIGID  RF00005_rep.0_M15347.1/1040-968_4            ORIGHEAD  RF00005_rep.0   
+CLUSTER  1  CM_SCORE  73.3  CMSEARCH  1.1  ORIGID  RF00005_rep.24_AC004941.2/32735-32806_20     ORIGHEAD  RF00005_rep.24  
+CLUSTER  1  CM_SCORE  69.4  CMSEARCH  1.1  ORIGID  RF00005_rep.23_M16479.1/42-123_19            ORIGHEAD  RF00005_rep.23  
+CLUSTER  1  CM_SCORE  69.3  CMSEARCH  1.1  ORIGID  RF00005_rep.4_Z98744.2/66305-66234_45        ORIGHEAD  RF00005_rep.4   
+CLUSTER  1  CM_SCORE  47.8  CMSEARCH  1.1  ORIGID  RF00005_rep.26_AF346999.1/4402-4331_22       ORIGHEAD  RF00005_rep.26  
+CLUSTER  1  CM_SCORE  45.6  CMSEARCH  1.1  ORIGID  RF00005_rep.19_AF134583.1/1816-1744_14       ORIGHEAD  RF00005_rep.19  
+CLUSTER  1  CM_SCORE  34.4  CMSEARCH  1.1  ORIGID  RF00005_rep.32_AF347015.1/5892-5827_29       ORIGHEAD  RF00005_rep.32  
--- a/test-data/RESULTS/2.cluster.all	Fri Dec 16 07:34:49 2016 -0500
+++ b/test-data/RESULTS/2.cluster.all	Thu Dec 22 09:06:48 2016 -0500
@@ -1,13 +1,13 @@
-CLUSTER  2  CM_SCORE  86.6  MODEL     1.2  RF00641_rep.2   
-CLUSTER  2  CM_SCORE  79.8  MODEL     1.2  RF00641_rep.4   
-CLUSTER  2  CM_SCORE  76.7  MODEL     1.2  RF00641_rep.12  
-CLUSTER  2  CM_SCORE  75.4  MODEL     1.2  RF00641_rep.9   
-CLUSTER  2  CM_SCORE  74.4  MODEL     1.2  RF00641_rep.11  
-CLUSTER  2  CM_SCORE  73.2  MODEL     1.2  RF00641_rep.5   
-CLUSTER  2  CM_SCORE  72.1  MODEL     1.2  RF00641_rep.7   
-CLUSTER  2  CM_SCORE  70.3  MODEL     1.2  RF00641_rep.1   
-CLUSTER  2  CM_SCORE  69.6  MODEL     1.2  RF00641_rep.6   
-CLUSTER  2  CM_SCORE  62.3  MODEL     1.2  RF00641_rep.8   
-CLUSTER  2  CM_SCORE  57.6  MODEL     1.2  RF00641_rep.13  
-CLUSTER  2  CM_SCORE  81.8  CMSEARCH  1.2  RF00641_rep.0   
-CLUSTER  2  CM_SCORE  56.6  CMSEARCH  1.2  RF00641_rep.10  
+CLUSTER  2  CM_SCORE  86.6  MODEL     1.2  ORIGID  RF00641_rep.2_AC208187.2/29806-29886_82        ORIGHEAD  RF00641_rep.2   
+CLUSTER  2  CM_SCORE  79.8  MODEL     1.2  ORIGID  RF00641_rep.4_ABBA01048514.1/248685-248607_84  ORIGHEAD  RF00641_rep.4   
+CLUSTER  2  CM_SCORE  76.7  MODEL     1.2  ORIGID  RF00641_rep.12_AL132709.5/56006-55928_79       ORIGHEAD  RF00641_rep.12  
+CLUSTER  2  CM_SCORE  75.4  MODEL     1.2  ORIGID  RF00641_rep.9_AL132709.5/98273-98194_89        ORIGHEAD  RF00641_rep.9   
+CLUSTER  2  CM_SCORE  74.4  MODEL     1.2  ORIGID  RF00641_rep.11_AC208187.2/25743-25822_78       ORIGHEAD  RF00641_rep.11  
+CLUSTER  2  CM_SCORE  73.2  MODEL     1.2  ORIGID  RF00641_rep.5_AL132709.5/73699-73621_85        ORIGHEAD  RF00641_rep.5   
+CLUSTER  2  CM_SCORE  72.1  MODEL     1.2  ORIGID  RF00641_rep.7_AL132709.5/54880-54803_87        ORIGHEAD  RF00641_rep.7   
+CLUSTER  2  CM_SCORE  70.3  MODEL     1.2  ORIGID  RF00641_rep.1_AL132709.5/86350-86274_81        ORIGHEAD  RF00641_rep.1   
+CLUSTER  2  CM_SCORE  69.6  MODEL     1.2  ORIGID  RF00641_rep.6_AADB02234051.1/94-172_86         ORIGHEAD  RF00641_rep.6   
+CLUSTER  2  CM_SCORE  62.3  MODEL     1.2  ORIGID  RF00641_rep.8_AL132709.5/61554-61474_88        ORIGHEAD  RF00641_rep.8   
+CLUSTER  2  CM_SCORE  57.6  MODEL     1.2  ORIGID  RF00641_rep.13_AL132709.5/55165-55087_80       ORIGHEAD  RF00641_rep.13  
+CLUSTER  2  CM_SCORE  81.8  CMSEARCH  1.2  ORIGID  RF00641_rep.0_AL132709.5/95865-95787_76        ORIGHEAD  RF00641_rep.0   
+CLUSTER  2  CM_SCORE  56.6  CMSEARCH  1.2  ORIGID  RF00641_rep.10_AADD01141100.1/2839-2915_77     ORIGHEAD  RF00641_rep.10  
--- a/test-data/RESULTS/partitions/final_partition.used_cmsearch	Fri Dec 16 07:34:49 2016 -0500
+++ b/test-data/RESULTS/partitions/final_partition.used_cmsearch	Thu Dec 22 09:06:48 2016 -0500
@@ -1,2 +1,2 @@
+1.2
 1.1
-1.2