Merge pull request DRL#82 from douglasgscofield/master

refine nodesDB handling and writing of default db
brittanymareeott · Feb 25, 2019 · 366f455 · 366f455
2 parents a14630e + af55ced
commit 366f455
Show file tree

Hide file tree

Showing 3 changed files with 26 additions and 11 deletions.
diff --git a/lib/BtIO.py b/lib/BtIO.py
@@ -611,7 +611,9 @@ def parseNodesDB(**kwargs):
     Parsing names.dmp and nodes.dmp into the 'nodes_db' dict of dicts that
     gets JSON'ed into blobtools/data/nodes_db.json if this file
     does not exist. Nodes_db.json is used if neither "--names" and "--nodes"
-    nor "--db" is specified.
+    nor "--db" is specified. If all three are specified and "--db" does not
+    exist, then write 'nodes_db' to file specified by "--db". If all three
+    are specified and "--db" exists, error out.
     '''
     nodesDB = {}
     names_f = kwargs['names']
@@ -624,7 +626,12 @@ def parseNodesDB(**kwargs):
             BtLog.error('0', names_f)
         if not isfile(nodes_f):
             BtLog.error('0', nodes_f)
-        print BtLog.status_d['3'] % (nodes_f, names_f)
+        if (nodesDB_f):
+            if isfile(nodesDB_f):
+                BtLog.error('47', nodesDB_f)
+            BtLog.status_d['27'] % (nodesDB_f, nodes_f, names_f)
+        else:
+            print BtLog.status_d['3'] % (nodes_f, names_f)
         try:
             nodesDB = readNamesNodes(names_f, nodes_f)
         except:
@@ -645,11 +652,17 @@ def parseNodesDB(**kwargs):
             nodesDB = readNodesDB(nodesDB_default)
         except:
             BtLog.error('27', nodesDB_default)
-        nodesDB_f = nodesDB_default
 
-    # Write nodesDB if not available
-    if not isfile(nodesDB_default):
-        writeNodesDB(nodesDB, nodesDB_default)
+    # Write nodesDB if names, nodes, nodesDB all given and nodesDB does not
+    # exist.  Otherwise, write to nodesDB_default if it does not exist, unless
+    # nodesDB given, then do nothing with nodesDB_default.
+    if (nodes_f and names_f and nodesDB_f):
+        print BtLog.status_d['28'] % nodesDB_f
+        writeNodesDB(nodesDB, nodesDB_f)
+    elif (not nodesDB_f and not isfile(nodesDB_default)):
+        nodesDB_f = nodesDB_default
+        print BtLog.status_d['5'] % nodesDB_f
+        writeNodesDB(nodesDB, nodesDB_f)
 
     return nodesDB, nodesDB_f
 
@@ -691,7 +704,6 @@ def readNodesDB(nodesDB_f):
     return nodesDB
 
 def writeNodesDB(nodesDB, nodesDB_f):
-    print BtLog.status_d['5'] % nodesDB_f
     nodes_count = nodesDB['nodes_count']
     i = 0
     with open(nodesDB_f, 'w') as fh:

diff --git a/lib/BtLog.py b/lib/BtLog.py
@@ -79,8 +79,8 @@ def progress(iteration, steps, max_value, no_limit=False):
     '43': '[ERROR:43] : %s could not be found.',
     '44': '[ERROR:44] : Please specify integers for --map_col_sseqid and --map_col_taxid.',
     '45': '[ERROR:45] : Both --min_score and --min_diff must be numbers.',
-    '46': '[ERROR:46] : Score in %s must be a float, not \'%s\'.'
-
+    '46': '[ERROR:46] : Score in %s must be a float, not \'%s\'.',
+    '47': '[ERROR:47] : Cannot create new "--db" file from "--names", "--nodes", "--db" file exists. %s'
 }
 
 warn_d = {
@@ -125,7 +125,9 @@ def progress(iteration, steps, max_value, no_limit=False):
     '23': '[+] Filtered %s (pairs=%s) ...',
     '24': '[+] Writing %s',
     '25': '[+] Gzip\'ing %s',
-    '26': '[+] Reading %s'
+    '26': '[+] Reading %s',
+    '27': '[+] Creating nodesDB %s from %s and %s',
+    '28': '[+] Store nodesDB in %s',
 }
 
 info_d = {

diff --git a/lib/create.py b/lib/create.py
@@ -32,7 +32,8 @@
                                         taxonomies have equal scores (otherwise "unresolved") [default: False]
         --nodes <NODES>                 NCBI nodes.dmp file. Not required if '--db'
         --names <NAMES>                 NCBI names.dmp file. Not required if '--db'
-        --db <NODESDB>                  NodesDB file (default: $BLOBTOOLS/data/nodesDB.txt).
+        --db <NODESDB>                  NodesDB file (default: $BLOBTOOLS/data/nodesDB.txt).  If --nodes, --names and --db
+                                        are all given and NODESDB does not exist, create it from NODES and NAMES.
         -b, --bam <BAM>...              BAM file(s), can be specified multiple times
         -s, --sam <SAM>...              SAM file(s), can be specified multiple times
         -a, --cas <CAS>...              CAS file(s) (requires clc_mapping_info in $PATH), can be specified multiple times