From ab4dc39a67ddecbd75fdf7090a2d773149f4baf5 Mon Sep 17 00:00:00 2001
From: Jim Johnson <jj@umn.edu>
Date: Sun, 5 Jul 2015 04:33:39 -0500
Subject: [PATCH] Add datatype: galaxy.datatypes.proteomics:MzSQlite extension:
 mz.sqlite This is a SQLite schema for representing Peptide Spectral Matches
 from mzid and mzML datatypes.

---
 config/datatypes_conf.xml.sample   |  2 ++
 lib/galaxy/datatypes/proteomics.py | 25 +++++++++++++++++++++++++
 2 files changed, 27 insertions(+)

diff --git a/config/datatypes_conf.xml.sample b/config/datatypes_conf.xml.sample
index 02ab894525b..da8c9ef7e0c 100644
--- a/config/datatypes_conf.xml.sample
+++ b/config/datatypes_conf.xml.sample
@@ -167,6 +167,7 @@
     <datatype extension="mzxml" type="galaxy.datatypes.proteomics:MzXML" mimetype="application/xml" display_in_upload="true" />
     <datatype extension="ms2" type="galaxy.datatypes.proteomics:Ms2" display_in_upload="true" />
     <datatype extension="mzq" type="galaxy.datatypes.proteomics:MzQuantML" mimetype="application/xml" display_in_upload="true" />
+    <datatype extension="mz.sqlite" type="galaxy.datatypes.proteomics:MzSQlite" mimetype="application/octet-stream" display_in_upload="true" />
     <datatype extension="traml" type="galaxy.datatypes.proteomics:TraML" mimetype="application/xml" display_in_upload="true" />
     <datatype extension="featurexml" type="galaxy.datatypes.proteomics:FeatureXML" mimetype="application/xml" display_in_upload="true" />
     <datatype extension="consensusxml" type="galaxy.datatypes.proteomics:ConsensusXML" mimetype="application/xml" display_in_upload="true" />
@@ -422,6 +423,7 @@
     <sniffer type="galaxy.datatypes.tabular:Vcf"/>
     <sniffer type="galaxy.datatypes.binary:TwoBit"/>
     <sniffer type="galaxy.datatypes.binary:GeminiSQLite"/>
+    <sniffer type="galaxy.datatypes.proteomics:MzSQlite"/>
     <sniffer type="galaxy.datatypes.binary:SQlite"/>
     <sniffer type="galaxy.datatypes.binary:Bam"/>
     <sniffer type="galaxy.datatypes.binary:Sff"/>
diff --git a/lib/galaxy/datatypes/proteomics.py b/lib/galaxy/datatypes/proteomics.py
index c505197c58d..020a35a9a5d 100644
--- a/lib/galaxy/datatypes/proteomics.py
+++ b/lib/galaxy/datatypes/proteomics.py
@@ -398,3 +398,28 @@ class XHunterAslFormat(Binary):
 class Sf3(Binary):
     """Class describing a Scaffold SF3 files"""
     file_ext = "sf3"
+
+class MzSQLite( SQlite ):
+    """Class describing a Proteomics Sqlite database """
+    file_ext = "mz.sqlite"
+
+    def set_meta( self, dataset, overwrite = True, **kwd ):
+        super( MzSQLite, self ).set_meta( dataset, overwrite = overwrite, **kwd )
+
+    def sniff( self, filename ):
+        if super( MzSQLite, self ).sniff( filename ):
+            mz_table_names = [ "DBSequence", "Modification", "Peaks", "Peptide", "PeptideEvidence", "Score", "SearchDatabase", "Source", "SpectraData", "Spectrum", "SpectrumIdentification]
+            try:
+                conn = sqlite.connect( filename )
+                c = conn.cursor()
+                tables_query = "SELECT name FROM sqlite_master WHERE type='table' ORDER BY name"
+                result = c.execute( tables_query ).fetchall()
+                result = map( lambda x: x[0], result )
+                for table_name in mz_table_names:
+                    if table_name not in result:
+                        return False
+                return True
+            except Exception, e:
+                log.warn( '%s, sniff Exception: %s', self, e )
+        return False
+