protk 1.3.1.pre3 → 1.4.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +21 -19
- data/bin/add_retention_times.rb +1 -1
- data/bin/interprophet.rb +16 -5
- data/bin/make_decoy.rb +1 -1
- data/bin/manage_db.rb +1 -1
- data/bin/mascot_search.rb +2 -2
- data/bin/mascot_to_pepxml.rb +1 -1
- data/bin/msgfplus_search.rb +26 -9
- data/bin/omssa_search.rb +1 -1
- data/bin/peptide_prophet.rb +57 -20
- data/bin/pepxml_to_table.rb +15 -2
- data/bin/protein_prophet.rb +41 -1
- data/bin/protk_setup.rb +2 -2
- data/bin/protxml_to_gff.rb +50 -42
- data/bin/protxml_to_psql.rb +1 -1
- data/bin/protxml_to_table.rb +16 -3
- data/bin/repair_run_summary.rb +1 -1
- data/bin/sixframe.rb +2 -2
- data/bin/swissprot_to_table.rb +1 -1
- data/bin/tandem_search.rb +1 -1
- data/bin/tandem_to_pepxml.rb +1 -1
- data/lib/protk/constants.rb +2 -1
- data/lib/protk/convert_util.rb +1 -1
- data/lib/protk/data/tandem-style.css +349 -0
- data/lib/protk/data/tandem-style.xsl +264 -0
- data/lib/protk/data/tandem_gpm_defaults.xml +3 -3
- data/lib/protk/data/tandem_isb_kscore_defaults.xml +2 -0
- data/lib/protk/data/tandem_isb_native_defaults.xml +3 -0
- data/lib/protk/data/tandem_params.xml +0 -8
- data/lib/protk/fastadb.rb +1 -1
- data/lib/protk/galaxy_stager.rb +14 -3
- data/lib/protk/galaxy_util.rb +39 -31
- data/lib/protk/gffdb.rb +6 -1
- data/lib/protk/manage_db_rakefile.rake +1 -1
- data/lib/protk/manage_db_tool.rb +1 -1
- data/lib/protk/pepxml.rb +159 -7
- data/lib/protk/plasmodb.rb +1 -1
- data/lib/protk/prophet_tool.rb +20 -52
- data/lib/protk/setup_rakefile.rake +18 -11
- data/lib/protk/tandem_search_tool.rb +20 -7
- data/lib/protk/tool.rb +1 -1
- data/lib/protk/uniprot_mapper.rb +1 -1
- metadata +10 -14
|
@@ -0,0 +1,264 @@
|
|
|
1
|
+
<?xml version="1.0" encoding="ISO-8859-1"?>
|
|
2
|
+
<xsl:stylesheet version="1.0"
|
|
3
|
+
xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
|
|
4
|
+
xmlns:GAML="http://www.bioml.com/gaml/" >
|
|
5
|
+
<!--
|
|
6
|
+
X! tandem default style sheet
|
|
7
|
+
Copyright (C) 2003-2004 Ronald C. Beavis
|
|
8
|
+
Version 2004.03.01
|
|
9
|
+
All Rights Reserved
|
|
10
|
+
This source code is distributed under the terms of the
|
|
11
|
+
Artistic License.
|
|
12
|
+
-->
|
|
13
|
+
<xsl:template match="/">
|
|
14
|
+
<html>
|
|
15
|
+
<head>
|
|
16
|
+
<title>thegpm site 000 <xsl:value-of select="/bioml/@label" /></title>
|
|
17
|
+
<link rel="stylesheet" href="/tandem/tandem-style.css" />
|
|
18
|
+
<SCRIPT lanuage="JavaScript">
|
|
19
|
+
|
|
20
|
+
function changeState(id)
|
|
21
|
+
{
|
|
22
|
+
var block = document.getElementById(id);
|
|
23
|
+
if(block.style.display == 'none') {
|
|
24
|
+
block.style.display = 'block';
|
|
25
|
+
}
|
|
26
|
+
else {
|
|
27
|
+
block.style.display = 'none';
|
|
28
|
+
}
|
|
29
|
+
}
|
|
30
|
+
</SCRIPT>
|
|
31
|
+
</head>
|
|
32
|
+
|
|
33
|
+
<body bgcolor="#FFFFFF">
|
|
34
|
+
<TABLE bgcolor="#d0d0d0" CELLSPACING="2" CELLPADDING="2">
|
|
35
|
+
<TR>
|
|
36
|
+
<TD WIDTH="700" VALIGN="TOP" ALIGN="LEFT" class="top_note">X! tandem results</TD>
|
|
37
|
+
</TR>
|
|
38
|
+
<TR>
|
|
39
|
+
<TD WIDTH="700" VALIGN="TOP" ALIGN="LEFT" class="top_note"><B><xsl:value-of select="/bioml/@label" /></B></TD>
|
|
40
|
+
</TR>
|
|
41
|
+
</TABLE>
|
|
42
|
+
<BR></BR>
|
|
43
|
+
<table border="1" bgcolor="#d0ffd0" cellpadding="2" cellspacing="2">
|
|
44
|
+
<xsl:apply-templates select="/bioml/group" />
|
|
45
|
+
</table>
|
|
46
|
+
<xsl:if test="not(/bioml/group)">
|
|
47
|
+
<TABLE CELLSPACING="2" CELLPADDING="2">
|
|
48
|
+
<TR>
|
|
49
|
+
<TD WIDTH="700" VALIGN="TOP" ALIGN="LEFT">No models were provided.</TD>
|
|
50
|
+
</TR>
|
|
51
|
+
</TABLE>
|
|
52
|
+
|
|
53
|
+
</xsl:if>
|
|
54
|
+
</body>
|
|
55
|
+
</html>
|
|
56
|
+
</xsl:template>
|
|
57
|
+
|
|
58
|
+
<xsl:template match="group">
|
|
59
|
+
<xsl:if test="not(contains(@label,'parameters'))">
|
|
60
|
+
<tr><td>
|
|
61
|
+
<DIV onClick="changeState('{generate-id()}');" class="e" title="click to see/hide sequences and evidence">
|
|
62
|
+
<SPAN CLASS="top_label">
|
|
63
|
+
#<xsl:value-of select="@id" />,
|
|
64
|
+
e = <xsl:value-of select="@expect" />,
|
|
65
|
+
M+H = <sup><xsl:value-of select="@z" /></sup><xsl:value-of select="@mh" />
|
|
66
|
+
<sup><xsl:value-of select="./protein/peptide/domain/@delta" /></sup>,
|
|
67
|
+
<SPAN class="aa_s"><xsl:value-of select="./protein/peptide/domain/@pre" /></SPAN>
|
|
68
|
+
<SUP><xsl:value-of select="./protein/peptide/domain/@start" /></SUP>
|
|
69
|
+
<SPAN class="aa_h"><xsl:value-of select="./protein/peptide/domain/@seq" /></SPAN>
|
|
70
|
+
<SUP><xsl:value-of select="./protein/peptide/domain/@end" /></SUP>
|
|
71
|
+
<SPAN class="aa_s"><xsl:value-of select="./protein/peptide/domain/@post" /></SPAN>,
|
|
72
|
+
<xsl:for-each select="./protein[1]/peptide[1]/domain[1]/aa">
|
|
73
|
+
|
|
74
|
+
<SUP><xsl:value-of select="@at" /></SUP><xsl:value-of select="@type" />(<xsl:value-of select="@modified" />),
|
|
75
|
+
|
|
76
|
+
</xsl:for-each>
|
|
77
|
+
</SPAN><BR></BR>
|
|
78
|
+
<SPAN CLASS="top_note">
|
|
79
|
+
<span class="b"> log(E) = <xsl:value-of select="./protein/@expect" /></span>,
|
|
80
|
+
<xsl:value-of select="@label" />
|
|
81
|
+
</SPAN>
|
|
82
|
+
</DIV>
|
|
83
|
+
|
|
84
|
+
<DIV id="{generate-id()}" STYLE="display:none" class="k">
|
|
85
|
+
<TABLE border="1" bgcolor="#ffd0ff" CELLPADDING="1" CELLSPACING="1">
|
|
86
|
+
<xsl:apply-templates select="protein" mode="sequence"/>
|
|
87
|
+
</TABLE>
|
|
88
|
+
<xsl:apply-templates select="group" mode="support"/>
|
|
89
|
+
</DIV>
|
|
90
|
+
</td></tr>
|
|
91
|
+
</xsl:if>
|
|
92
|
+
|
|
93
|
+
<xsl:if test="contains(@label,'parameters')">
|
|
94
|
+
<DIV onClick="changeState('{generate-id()}');" title="click to see/hide values" class="e">
|
|
95
|
+
<SPAN CLASS="top_label">+ <xsl:value-of select="@label" />
|
|
96
|
+
</SPAN>
|
|
97
|
+
</DIV>
|
|
98
|
+
<DIV id="{generate-id()}" class="k" STYLE="display:none">
|
|
99
|
+
<BR></BR>
|
|
100
|
+
<TABLE border="0" bgcolor="#ffd0d0" CELLPADDING="1" CELLSPACING="1">
|
|
101
|
+
<SPAN CLASS="top_note"><xsl:apply-templates select="note" />
|
|
102
|
+
</SPAN>
|
|
103
|
+
</TABLE>
|
|
104
|
+
<HR></HR>
|
|
105
|
+
</DIV>
|
|
106
|
+
</xsl:if>
|
|
107
|
+
</xsl:template>
|
|
108
|
+
|
|
109
|
+
<xsl:template match="protein" mode="description">
|
|
110
|
+
<BR></BR><SPAN CLASS="top_note"><B><xsl:value-of select="@id" /></B> : </SPAN><PRE><xsl:apply-templates select="note" /></PRE>
|
|
111
|
+
</xsl:template>
|
|
112
|
+
|
|
113
|
+
<xsl:template match="protein" mode="sequence">
|
|
114
|
+
<tr><td>
|
|
115
|
+
<DIV onClick="changeState('{generate-id()}');" title="click to see/hide details">
|
|
116
|
+
<SPAN CLASS="top_label"><xsl:value-of select="@id" />: <xsl:value-of select="@label" />
|
|
117
|
+
<xsl:apply-templates select="file" /></SPAN>
|
|
118
|
+
</DIV>
|
|
119
|
+
<DIV id="{generate-id()}" class="k" STYLE="display:none">
|
|
120
|
+
<xsl:apply-templates select="peptide" />
|
|
121
|
+
</DIV>
|
|
122
|
+
</td></tr>
|
|
123
|
+
</xsl:template>
|
|
124
|
+
|
|
125
|
+
<xsl:template match="file">
|
|
126
|
+
<SPAN class="top_label">(<xsl:value-of select="@URL" />)</SPAN>
|
|
127
|
+
</xsl:template>
|
|
128
|
+
|
|
129
|
+
<xsl:template match="aa">
|
|
130
|
+
<SPAN class="top_note"><SUP><xsl:value-of select="@at" /></SUP><xsl:value-of select="@type" />(<xsl:value-of select="@modified" />),</SPAN>
|
|
131
|
+
</xsl:template>
|
|
132
|
+
|
|
133
|
+
<xsl:template match="GAML:attribute">
|
|
134
|
+
<SPAN class="small_label"><xsl:value-of select="@type" /> = <xsl:value-of select="text()" />, </SPAN>
|
|
135
|
+
</xsl:template>
|
|
136
|
+
|
|
137
|
+
<xsl:template match="GAML:trace">
|
|
138
|
+
|
|
139
|
+
<SPAN class="small_label">
|
|
140
|
+
<B>
|
|
141
|
+
<xsl:value-of select="@type" />
|
|
142
|
+
</B>
|
|
143
|
+
</SPAN>
|
|
144
|
+
<TABLE BORDER="0" bgcolor="#ffffd0" CELLPADDING="1" CELLSPACING="1">
|
|
145
|
+
<xsl:if test="GAML:attribute">
|
|
146
|
+
<TR>
|
|
147
|
+
<TD WIDTH="500" VALIGN="TOP" ALIGN="LEFT">
|
|
148
|
+
parameters: <xsl:apply-templates select="GAML:attribute" />
|
|
149
|
+
</TD>
|
|
150
|
+
</TR>
|
|
151
|
+
</xsl:if>
|
|
152
|
+
<TR>
|
|
153
|
+
<TD WIDTH="500" VALIGN="TOP" ALIGN="LEFT">
|
|
154
|
+
x-values
|
|
155
|
+
</TD>
|
|
156
|
+
</TR>
|
|
157
|
+
<TR>
|
|
158
|
+
<TD WIDTH="500" VALIGN="TOP" ALIGN="LEFT">
|
|
159
|
+
<xsl:value-of select="./GAML:Xdata/GAML:values/text()" />
|
|
160
|
+
</TD>
|
|
161
|
+
</TR>
|
|
162
|
+
<TR>
|
|
163
|
+
<TD WIDTH="500" VALIGN="TOP" ALIGN="LEFT">
|
|
164
|
+
y-values
|
|
165
|
+
</TD>
|
|
166
|
+
</TR>
|
|
167
|
+
<TR>
|
|
168
|
+
<TD WIDTH="500" VALIGN="TOP" ALIGN="LEFT">
|
|
169
|
+
<xsl:value-of select="./GAML:Ydata/GAML:values/text()" />
|
|
170
|
+
</TD>
|
|
171
|
+
</TR>
|
|
172
|
+
</TABLE>
|
|
173
|
+
</xsl:template>
|
|
174
|
+
|
|
175
|
+
<xsl:template match="group" mode="support">
|
|
176
|
+
<xsl:if test="@type='support'">
|
|
177
|
+
<DIV onClick="changeState('{generate-id()}');" class="e" title="click to see/hide evidence">
|
|
178
|
+
<SPAN CLASS="top_label">
|
|
179
|
+
<xsl:value-of select="@label" />
|
|
180
|
+
</SPAN></DIV>
|
|
181
|
+
<DIV id="{generate-id()}" STYLE="display:none" class="k">
|
|
182
|
+
<xsl:apply-templates select="GAML:trace" />
|
|
183
|
+
</DIV>
|
|
184
|
+
</xsl:if>
|
|
185
|
+
</xsl:template>
|
|
186
|
+
|
|
187
|
+
|
|
188
|
+
|
|
189
|
+
<xsl:template match="peptide">
|
|
190
|
+
<xsl:apply-templates select="domain" />
|
|
191
|
+
<DIV onClick="changeState('{generate-id()}');" class="e" title="click to see/hide sequence">
|
|
192
|
+
<SPAN class="top_label">
|
|
193
|
+
log(e) = <xsl:value-of select="./../@expect" />,
|
|
194
|
+
<xsl:value-of select="./../@label" />
|
|
195
|
+
</SPAN>
|
|
196
|
+
</DIV>
|
|
197
|
+
<DIV id="{generate-id()}" STYLE="display:none">
|
|
198
|
+
<TABLE BORDER="0" gbcolor="#d0d0ff">
|
|
199
|
+
<TR>
|
|
200
|
+
<TD WIDTH="500" ALIGN="LEFT" VALIGN="TOP" CLASS="residues"><xsl:value-of select="text()" />
|
|
201
|
+
</TD>
|
|
202
|
+
</TR>
|
|
203
|
+
</TABLE>
|
|
204
|
+
</DIV>
|
|
205
|
+
</xsl:template>
|
|
206
|
+
|
|
207
|
+
<xsl:template match="domain">
|
|
208
|
+
<SPAN CLASS="top_note">
|
|
209
|
+
<B><xsl:value-of select="@id" /></B>:
|
|
210
|
+
e = <xsl:value-of select="@expect" />,
|
|
211
|
+
<SUP><xsl:value-of select="@start" /></SUP><SPAN class="aa_h"><xsl:value-of select="@seq" /></SPAN><SUP><xsl:value-of select="@end" /></SUP>,
|
|
212
|
+
<xsl:apply-templates select="aa"/><BR></BR>
|
|
213
|
+
M+H = <xsl:value-of select="@mh" /> Da,
|
|
214
|
+
<SPAN CLASS="greek">d</SPAN> = <xsl:value-of select="@delta" />,
|
|
215
|
+
!score = <xsl:value-of select="@hyperscore" />,
|
|
216
|
+
y/b: scores = <xsl:value-of select="@y_score" />/<xsl:value-of select="@b_score" />,
|
|
217
|
+
ions = <xsl:value-of select="@y_ions" />/<xsl:value-of select="@b_ions" />
|
|
218
|
+
</SPAN>
|
|
219
|
+
<BR></BR>
|
|
220
|
+
</xsl:template>
|
|
221
|
+
|
|
222
|
+
<xsl:template match="note">
|
|
223
|
+
<xsl:if test="not(contains(@label,'description'))">
|
|
224
|
+
<TR>
|
|
225
|
+
<TD WIDTH="350" ALIGN="RIGHT"><xsl:value-of select="@label" /> = </TD>
|
|
226
|
+
<TD WIDTH="350" ALIGN="LEFT"><xsl:value-of select="text()" /></TD>
|
|
227
|
+
</TR>
|
|
228
|
+
</xsl:if>
|
|
229
|
+
<xsl:if test="contains(@label,'description')">
|
|
230
|
+
<SPAN CLASS="top_note">
|
|
231
|
+
<xsl:choose>
|
|
232
|
+
<xsl:when test="contains(self::note,'ENSMUSP')">
|
|
233
|
+
<a target="_BLANK">
|
|
234
|
+
<xsl:attribute name="href">http://www.ensembl.org/Mus_musculus/protview?peptide=<xsl:value-of select="text()"/></xsl:attribute>
|
|
235
|
+
<xsl:attribute name="title">View Ensembl Protein Report</xsl:attribute>
|
|
236
|
+
<span class="ensembl"><xsl:value-of select="text()" /></span>
|
|
237
|
+
</a>
|
|
238
|
+
</xsl:when>
|
|
239
|
+
<xsl:when test="contains(self::note,'ENSRNOP')">
|
|
240
|
+
<a target="_BLANK">
|
|
241
|
+
<xsl:attribute name="href">http://www.ensembl.org/Rattus_norvegicus/protview?peptide=<xsl:value-of select="text()"/></xsl:attribute>
|
|
242
|
+
<xsl:attribute name="title">View Ensembl Protein Report</xsl:attribute>
|
|
243
|
+
<span class="ensembl"><xsl:value-of select="text()" /></span>
|
|
244
|
+
</a>
|
|
245
|
+
</xsl:when>
|
|
246
|
+
<xsl:when test="contains(self::note,'ENSP')">
|
|
247
|
+
<a target="_BLANK">
|
|
248
|
+
<xsl:attribute name="href">http://www.ensembl.org/Homo_sapiens/protview?peptide=<xsl:value-of select="text()"/></xsl:attribute>
|
|
249
|
+
<xsl:attribute name="title">View Ensembl Protein Report</xsl:attribute>
|
|
250
|
+
<span class="ensembl"><xsl:value-of select="text()" /></span>
|
|
251
|
+
</a>
|
|
252
|
+
</xsl:when>
|
|
253
|
+
<xsl:otherwise>
|
|
254
|
+
<xsl:value-of select="text()" />
|
|
255
|
+
</xsl:otherwise>
|
|
256
|
+
</xsl:choose>
|
|
257
|
+
</SPAN>
|
|
258
|
+
</xsl:if>
|
|
259
|
+
</xsl:template>
|
|
260
|
+
|
|
261
|
+
|
|
262
|
+
|
|
263
|
+
</xsl:stylesheet>
|
|
264
|
+
|
|
@@ -10,11 +10,11 @@
|
|
|
10
10
|
<note type="input" label="spectrum, fragment monoisotopic mass error units">Daltons</note>
|
|
11
11
|
<note>The value for this parameter may be 'Daltons' or 'ppm': all other values are ignored</note>
|
|
12
12
|
<note type="input" label="spectrum, parent monoisotopic mass error units">ppm</note>
|
|
13
|
-
|
|
13
|
+
<note>The value for this parameter may be 'Daltons' or 'ppm': all other values are ignored</note>
|
|
14
14
|
<note type="input" label="spectrum, fragment mass type">monoisotopic</note>
|
|
15
|
-
|
|
15
|
+
<note>values are monoisotopic|average </note>
|
|
16
16
|
|
|
17
|
-
<note>spectrum conditioning parameters</note>
|
|
17
|
+
<note>spectrum conditioning parameters</note>
|
|
18
18
|
<note type="input" label="spectrum, dynamic range">100.0</note>
|
|
19
19
|
<note>The peaks read in are normalized so that the most intense peak
|
|
20
20
|
is set to the dynamic range value. All peaks with values of less that
|
|
@@ -3,6 +3,9 @@
|
|
|
3
3
|
<bioml>
|
|
4
4
|
|
|
5
5
|
<note>spectrum parameters</note>
|
|
6
|
+
<note type="input" label="spectrum, parent monoisotopic mass error minus">2.0</note>
|
|
7
|
+
<note type="input" label="spectrum, parent monoisotopic mass error plus">4.0</note>
|
|
8
|
+
<note>PRECURSOR MASS TOLERANCE. This is monoisotopic mass, so for non-accurate-mass instruments, for which the precursor is often taken nearer to the isotopically averaged mass, an asymmetric tolerance (-2.0 Da to 4.0 Da) is preferable. This somewhat imitates a (-3.0 Da to 3.0 Da) window for averaged mass (but not exactly).</note>
|
|
6
9
|
<note type="input" label="spectrum, parent monoisotopic mass isotope error">no</note>
|
|
7
10
|
<note type="input" label="spectrum, parent monoisotopic mass error units">Daltons</note>
|
|
8
11
|
<note>The value for this parameter may be 'Daltons' or 'ppm': all other values are ignored</note>
|
|
@@ -15,14 +15,6 @@
|
|
|
15
15
|
</note>
|
|
16
16
|
<note type="input" label="list path, default parameters">no default</note>
|
|
17
17
|
|
|
18
|
-
<note> MODIFICATIONS. No Default</note>
|
|
19
|
-
|
|
20
|
-
<note type="input" label="protein, cleavage C-terminal mass change"></note>
|
|
21
|
-
<note type="input" label="protein, cleavage N-terminal mass change"></note>
|
|
22
|
-
<note type="input" label="protein, N-terminal residue modification mass"></note>
|
|
23
|
-
<note type="input" label="protein, C-terminal residue modification mass"></note>
|
|
24
|
-
<note> These are *static* modifications on the PROTEINS' N or C-termini. </note>
|
|
25
|
-
|
|
26
18
|
|
|
27
19
|
<note>
|
|
28
20
|
REFINEMENT. Not supported via tandem_search.rb.
|
data/lib/protk/fastadb.rb
CHANGED
data/lib/protk/galaxy_stager.rb
CHANGED
|
@@ -2,17 +2,28 @@ $VERBOSE=nil
|
|
|
2
2
|
|
|
3
3
|
require 'pathname'
|
|
4
4
|
|
|
5
|
+
# A GalaxyStager object represents a single file whose path needs to be staged for use within galaxy
|
|
6
|
+
#
|
|
7
|
+
# Staging occurs on init and generally involves creating a symlink from the original path to the staged path
|
|
8
|
+
#
|
|
9
|
+
# By keeping GalaxyStager objects around until script completion it is possible to perform a
|
|
10
|
+
# restore_references operation in script output files.
|
|
11
|
+
#
|
|
5
12
|
class GalaxyStager
|
|
6
13
|
attr_accessor :staged_path
|
|
14
|
+
attr_accessor :staged_base_path
|
|
7
15
|
|
|
16
|
+
# Initialize and perform staging
|
|
17
|
+
#
|
|
8
18
|
def initialize(original_path, options = {})
|
|
9
19
|
options = { :name => nil, :extension => '', :force_copy => false }.merge(options)
|
|
20
|
+
|
|
10
21
|
@extension = options[:extension]
|
|
11
22
|
@original_path = Pathname.new(original_path)
|
|
12
23
|
@wd = Dir.pwd
|
|
13
24
|
@staged_name = options[:name] || @original_path.basename
|
|
14
|
-
@
|
|
15
|
-
@staged_path = "#{@
|
|
25
|
+
@staged_base_path = File.join(@wd, @staged_name)
|
|
26
|
+
@staged_path = "#{@staged_base_path}#{@extension}"
|
|
16
27
|
if options[:force_copy]
|
|
17
28
|
FileUtils.copy(@original_path, @staged_path)
|
|
18
29
|
else
|
|
@@ -20,7 +31,7 @@ class GalaxyStager
|
|
|
20
31
|
end
|
|
21
32
|
end
|
|
22
33
|
|
|
23
|
-
def replace_references(in_file)
|
|
34
|
+
def replace_references(in_file,replacement)
|
|
24
35
|
GalaxyStager.replace_references(in_file, @original_path, replacement)
|
|
25
36
|
end
|
|
26
37
|
|
data/lib/protk/galaxy_util.rb
CHANGED
|
@@ -12,40 +12,48 @@ class GalaxyUtil
|
|
|
12
12
|
fg
|
|
13
13
|
end
|
|
14
14
|
|
|
15
|
-
def self.stage_pepxml(input_pepxml_path)
|
|
16
|
-
|
|
17
|
-
|
|
15
|
+
def self.stage_pepxml(input_pepxml_path,options={})
|
|
16
|
+
options = { :extension => '.pep.xml', :force_copy => false }.merge(options)
|
|
17
|
+
GalaxyStager.new(input_pepxml_path, options )
|
|
18
18
|
end
|
|
19
19
|
|
|
20
|
-
def self.
|
|
21
|
-
|
|
22
|
-
|
|
23
|
-
# symbolic links to the peak lists with the correct extension and
|
|
24
|
-
# and indexes them if needed (both seem required for TPP quant
|
|
25
|
-
# tools) and then produces new protxml and pepxml files with paths
|
|
26
|
-
# updated to these new peak list files.
|
|
27
|
-
|
|
28
|
-
protxml_path="interact.prot.xml"
|
|
29
|
-
FileUtils.copy(input_protxml_path, "interact.prot.xml")
|
|
30
|
-
|
|
31
|
-
protxml = ProtXML.new(protxml_path)
|
|
32
|
-
pepxml_path = protxml.find_pep_xml()
|
|
33
|
-
|
|
34
|
-
protxml_stager = GalaxyStager.new(protxml_path, :extension => ".prot.xml", :force_copy => true)
|
|
35
|
-
pepxml_stager = GalaxyStager.new(pepxml_path, :name => "interact", :extension => ".xml", :force_copy => true)
|
|
36
|
-
pepxml_path = pepxml_stager.staged_path
|
|
37
|
-
pepxml_stager.replace_references(protxml_path)
|
|
38
|
-
runs = PepXML.new(pepxml_stager.staged_path).find_runs()
|
|
39
|
-
|
|
40
|
-
run_stagers = runs.map do |base_name, run|
|
|
41
|
-
run_stager = GalaxyStager.new(base_name, :extension => ".#{run[:type]}")
|
|
42
|
-
ConvertUtil.ensure_mzml_indexed(run_stager.staged_path)
|
|
43
|
-
run_stager.replace_references(pepxml_path, :base_only => true)
|
|
44
|
-
run_stager
|
|
45
|
-
end
|
|
46
|
-
|
|
47
|
-
protxml_path
|
|
20
|
+
def self.stage_fasta(input_path,options={})
|
|
21
|
+
options = { :extension => '.fasta', :force_copy => false }.merge(options)
|
|
22
|
+
GalaxyStager.new(input_path, options )
|
|
48
23
|
end
|
|
24
|
+
|
|
25
|
+
|
|
26
|
+
# Unused
|
|
27
|
+
|
|
28
|
+
# def self.stage_protxml(input_protxml_path)
|
|
29
|
+
# # This method takes in the path to a protxml created in Galaxy,
|
|
30
|
+
# # finds the dependent pepxml and peak lists (mzml files), creates
|
|
31
|
+
# # symbolic links to the peak lists with the correct extension and
|
|
32
|
+
# # and indexes them if needed (both seem required for TPP quant
|
|
33
|
+
# # tools) and then produces new protxml and pepxml files with paths
|
|
34
|
+
# # updated to these new peak list files.
|
|
35
|
+
|
|
36
|
+
# protxml_path="interact.prot.xml"
|
|
37
|
+
# FileUtils.copy(input_protxml_path, "interact.prot.xml")
|
|
38
|
+
|
|
39
|
+
# protxml = ProtXML.new(protxml_path)
|
|
40
|
+
# pepxml_path = protxml.find_pep_xml()
|
|
41
|
+
|
|
42
|
+
# protxml_stager = GalaxyStager.new(protxml_path, :extension => ".prot.xml", :force_copy => true)
|
|
43
|
+
# pepxml_stager = GalaxyStager.new(pepxml_path, :name => "interact", :extension => ".xml", :force_copy => true)
|
|
44
|
+
# pepxml_path = pepxml_stager.staged_path
|
|
45
|
+
# pepxml_stager.replace_references(protxml_path)
|
|
46
|
+
# runs = PepXML.new(pepxml_stager.staged_path).find_runs()
|
|
47
|
+
|
|
48
|
+
# run_stagers = runs.map do |base_name, run|
|
|
49
|
+
# run_stager = GalaxyStager.new(base_name, :extension => ".#{run[:type]}")
|
|
50
|
+
# ConvertUtil.ensure_mzml_indexed(run_stager.staged_path)
|
|
51
|
+
# run_stager.replace_references(pepxml_path, :base_only => true)
|
|
52
|
+
# run_stager
|
|
53
|
+
# end
|
|
54
|
+
|
|
55
|
+
# protxml_path
|
|
56
|
+
# end
|
|
49
57
|
|
|
50
58
|
|
|
51
59
|
|
data/lib/protk/gffdb.rb
CHANGED
|
@@ -7,7 +7,6 @@ class GFFDB
|
|
|
7
7
|
attr_accessor :id_to_records_map
|
|
8
8
|
|
|
9
9
|
def initialize(gff_file_path)
|
|
10
|
-
env = Constants.new
|
|
11
10
|
@database = gff_file_path
|
|
12
11
|
@id_to_records_map={}
|
|
13
12
|
@id_to_cds_map={}
|
|
@@ -29,9 +28,15 @@ class GFFDB
|
|
|
29
28
|
|
|
30
29
|
|
|
31
30
|
def make_index(input_gff)
|
|
31
|
+
env=Constants.instance
|
|
32
|
+
|
|
32
33
|
io = File.open(input_gff, "r")
|
|
34
|
+
env.log "Parsing Input GFF",:info
|
|
33
35
|
gffdb = Bio::GFF::GFF3.new(io) #parses the entire db
|
|
34
36
|
|
|
37
|
+
num_records = gffdb.records.length
|
|
38
|
+
env.log "Indexing #{num_records}", :info
|
|
39
|
+
|
|
35
40
|
# Now create the mapping from ids to records
|
|
36
41
|
gffdb.records.each do |record|
|
|
37
42
|
|