Mercurial > repos > jeremyjliu > region_motif_data_manager

changeset 10:a5421f83f972 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author jeremyjliu
date Thu, 26 Feb 2015 22:52:42 -0500
parents e4229c66d78d
children 24a2c2783fb8
files data_manager/data_manager_fetch_motifs.py data_manager/data_manager_fetch_motifs.xml tool-data/motif_databases.loc.sample
diffstat 3 files changed, 24 insertions(+), 9 deletions(-) [+]
line wrap: on
line diff
--- a/data_manager/data_manager_fetch_motifs.py	Sun Feb 01 15:37:11 2015 -0500
+++ b/data_manager/data_manager_fetch_motifs.py	Thu Feb 26 22:52:42 2015 -0500
@@ -23,26 +23,35 @@
 def download_motif_databases( data_manager_dict, params, target_directory, motif_db ):
 
     # Select download URL, file name, data table name, and path using motif_db selector variable
-    if motif_db == "pouya":
+    if motif_db == "encode":
         BGZ = ['http://compbio.med.harvard.edu/motif-enrichment/pouya_motifs.bed.bgz',
-                "pouya_motifs.bed.bgz", "pouya_bgz", "Pouya Encode Motifs (hg19) BGZ"]
+                "pouya_motifs.bed.bgz", "encode_bgz", "Encode Motifs (hg19) BGZ"]
         TBI = ['http://compbio.med.harvard.edu/motif-enrichment/pouya_motifs.bed.bgz.tbi',
-                "pouya_motifs.bed.bgz.tbi", "pouya_tbi", "Pouya Encode Motifs (hg19) TBI"]
+                "pouya_motifs.bed.bgz.tbi", "encode_tbi", "Encode Motifs (hg19) TBI"]
+        PWM = ['http://compbio.med.harvard.edu/motif-enrichment/pwms/pouya.pwms.from.seq.RData',
+                "pouya.pwms.from.seq.RData", "encode_pwm", "Encode Motifs (hg19) PWM"]        
     elif motif_db == "jaspar":
         BGZ = ['http://compbio.med.harvard.edu/motif-enrichment/jaspar_jolma_motifs.bed.bgz',
                 "jaspar_jolma_motifs.bed.bgz", "jaspar_bgz", "Jaspar and Jolma Motifs (hg19) BGZ"]
         TBI = ['http://compbio.med.harvard.edu/motif-enrichment/jaspar_jolma_motifs.bed.bgz.tbi',
                 "jaspar_jolma_motifs.bed.bgz.tbi", "jaspar_tbi", "Jaspar and Jolma Motifs (hg19) TBI"]
+        PWM = ['http://compbio.med.harvard.edu/motif-enrichment/pwms/jaspar.jolma.pwms.from.seq.RData',
+                "jaspar.jolma.pwms.from.seq.RData", "jaspar_pwm", "Jaspar and Jolma Motifs (hg19) PWM"]
     elif motif_db == "mouse":
         BGZ = ['http://compbio.med.harvard.edu/motif-enrichment/mm9_motifs_split.bed.bgz',
                 "mm9_motifs_split.bed.bgz", "mouse_bgz", "Mouse Motifs (mm9) BGZ"]
         TBI = ['http://compbio.med.harvard.edu/motif-enrichment/mm9_motifs_split.bed.bgz.tbi',
                 "mm9_motifs_split.bed.bgz.tbi", "mouse_tbi", "Mouse Motifs (mm9) TBI"]
+        PWM = ['http://compbio.med.harvard.edu/motif-enrichment/pwms/mm9.pwms.from.seq.RData',
+                "mm9.pwms.from.seq.RData", "mouse_pwm", "Mouse Motifs (mm9) PWM"]
     else:
         BGZ = ['http://compbio.med.harvard.edu/motif-enrichment/pouya_test_motifs.bed.bgz', 
-               "pouya_test_motifs.bed.bgz", "test_bgz", "Test Pouya Subset (hg19) BGZ"]
+               "pouya_test_motifs.bed.bgz", "test_bgz", "Test Encode Motifs (hg19) BGZ"]
         TBI = ['http://compbio.med.harvard.edu/motif-enrichment/pouya_test_motifs.bed.bgz.tbi',
-               "pouya_test_motifs.bed.bgz.tbi", "test_tbi", "Test Pouya Subset (hg19) TBI"]
+               "pouya_test_motifs.bed.bgz.tbi", "test_tbi", "Test Encode Motifs (hg19) TBI"]
+        PWM = ['http://compbio.med.harvard.edu/motif-enrichment/pwms/pouya.pwms.from.seq.RData',
+                "pouya.pwms.from.seq.RData", "test_pwm", "Test Encode Motifs (hg19) PWM"] 
+
 
     # Save and add motif bgz file to motif_databases data table
     bgz_reader = urllib2.urlopen( BGZ[0] )
@@ -56,6 +65,12 @@
                             TBI[1], TBI[2], TBI[3] )
     _add_data_table_entry( data_manager_dict, 'motif_databases', tbi_data_table_entry )
 
+    # Save and add motif pwm file to motif_databases data table
+    tbi_reader = urllib2.urlopen( PWM[0] )
+    tbi_data_table_entry = _stream_fasta_to_file( tbi_reader, target_directory, params,
+                            PWM[1], PWM[2], PWM[3] )
+    _add_data_table_entry( data_manager_dict, 'motif_databases', tbi_data_table_entry )
+
 def _add_data_table_entry( data_manager_dict, data_table, data_table_entry ):
     data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
     data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get( data_table, [] )
--- a/data_manager/data_manager_fetch_motifs.xml	Sun Feb 01 15:37:11 2015 -0500
+++ b/data_manager/data_manager_fetch_motifs.xml	Thu Feb 26 22:52:42 2015 -0500
@@ -3,8 +3,8 @@
     <command interpreter="python">data_manager_fetch_motifs.py "${out_file}" --motif_db ${motif_db_selector}</command>
     <inputs>
         <param name="motif_db_selector" type="select" label="Choose the source for the motif db">
-            <option value="test" selected="True">Test Pouya Subset (hg19)</option>
-            <option value="pouya">Pouya Encode Motifs (hg19)</option>
+            <option value="test" selected="True">Test Encode Motifs (hg19)</option>
+            <option value="encode">Encode Motifs (hg19)</option>
             <option value="jaspar">Jaspar and Jolma Motifs (hg19)</option>
             <option value="mouse">Mouse Motifs (mm9)</option>
         </param>
--- a/tool-data/motif_databases.loc.sample	Sun Feb 01 15:37:11 2015 -0500
+++ b/tool-data/motif_databases.loc.sample	Thu Feb 26 22:52:42 2015 -0500
@@ -5,5 +5,5 @@
 #
 # So, motif_databases.loc could look something like this:
 #
-# test_bgz	  Test Pouya Subset (hg19) BGZ	  /<path_to_galaxy>/tool-data/motifs/pouya_test_motifs.bed.bgz
-# test_tbi    Test Pouya Subset (hg19) TBI    /<path_to_galaxy>/tool-data/motifs/pouya_test_motifs.bed.bgz.tbi
+# test_bgz	  Test Encode Motifs (hg19) BGZ	  /<path_to_galaxy>/tool-data/motifs/pouya_test_motifs.bed.bgz
+# test_tbi    Test Encode motifs (hg19) TBI    /<path_to_galaxy>/tool-data/motifs/pouya_test_motifs.bed.bgz.tbi