# HG changeset patch
# User chrisb
# Date 1462536348 14400
# Node ID 0a5e0df170546447aa4845a476d1ce5ee66df1cd
# Parent 89592faa287574c7608d1636b91c39592f73c6db
Uploaded
diff -r 89592faa2875 -r 0a5e0df17054 get_data/cfg_array/.README_CFG_array.md.swp
Binary file get_data/cfg_array/.README_CFG_array.md.swp has changed
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/.idea/kegg_glycan.iml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/.idea/kegg_glycan.iml Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,12 @@
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/.idea/misc.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/.idea/misc.xml Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,14 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/.idea/modules.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/.idea/modules.xml Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,8 @@
+
+
+
+
+
+
+
+
\ No newline at end of file
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/.idea/workspace.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/.idea/workspace.xml Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,477 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ true
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ 1462439789911
+
+
+ 1462439789911
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/2
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/2 Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,54 @@
+
+from KEGG searches as follows should work
+
+
+/find/genes/shiga+toxin for keywords "shiga" and "toxin"
+/find/genes/"shiga toxin" for keywords "shiga toxin"
+
+these do not function on all databases esp. enzyme
+
+http://rest.kegg.jp/find/enzyme/deoxy+1.1.1 AND works
+http://rest.kegg.jp/find/enzyme/"deoxy%201.1.1" OR does not work
+http://rest.kegg.jp/find/enzyme/"deoxy 1.1.1" OR does not work
+
+0
+http://rest.kegg.jp/find/enzyme/ec:2.4.99.1+ec:2.4.99.6 AND works but return nothing as no overlap
+http://rest.kegg.jp/find/enzyme/"ec:2.4.99.1 ec:2.4.99.6" OR doesn't work
+http://rest.kegg.jp/find/enzyme/ec:2.4.99. works
+
+ ../../../virtualpy/bin/python findKEGG.py -d enzyme -q 2.4.99.1+2.4.99.6 . works but return None in python. which breaks file writing. rather fix this.
+
+
+
+
+
+# updated notes on KEGG behaviour
+
+True for release Release 78.0+/05-05, May 16 (see http://rest.kegg.jp/info/genes)
+
+multiple complex queries are not recommended.
+
+Using the shiga toxin example from http://www.genome.jp/kegg/rest/keggapi.html
+/find/genes/shiga+toxin for keywords "shiga" and "toxin"
+/find/genes/"shiga toxin" for keywords "shiga toxin"
+
+## AND example
+/find/genes/shiga+toxin for keywords "shiga" and "toxin"
+http://rest.kegg.jp/find/genes/shiga+toxin
+
+## OR example
+/find/genes/"shiga toxin" for keywords "shiga toxin"
+http://rest.kegg.jp/find/genes/"shiga%20toxin"
+
+## BROKEN AND
+http://rest.kegg.jp/find/genes/"shiga+toxin"
+If query is in put in quotes.
+ this result is the same as OR.
+
+## BROKEN OR
+http://rest.kegg.jp/find/genes/shiga%20toxin
+If query is not placed in quotations.
+ this result is the same as AND.
+
+# Recommendation
+Do not combine AND with OR queries
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/KEGGnotes.md
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/KEGGnotes.md Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,51 @@
+
+# notes on KEGG behaviour
+
+Tested with Release 78.0+/05-05, May 16 (see http://rest.kegg.jp/info/genes)
+
+*Recommendation 1 - Do not combine AND with OR queries*
+multiple complex queries are not recommended.
+
+*OR's on certain db's e.g. glycan, enzyme do not always return correctly*
+
+## Shiga toxin behaviour test
+Using the shiga toxin example from http://www.genome.jp/kegg/rest/keggapi.html
+/find/genes/shiga+toxin for keywords "shiga" and "toxin"
+/find/genes/"shiga toxin" for keywords "shiga toxin"
+
+results in test-data
+ - test-data/shiga-toxin-example-AND
+ - test-data/shiga-toxin-example-AND-broken
+ - test-data/shiga-toxin-example-OR
+ - test-data/shiga-toxin-example-OR-broken
+
+
+### AND example
+/find/genes/shiga+toxin for keywords "shiga" and "toxin"
+http://rest.kegg.jp/find/genes/shiga+toxin
+
+### OR example
+/find/genes/"shiga toxin" for keywords "shiga toxin"
+http://rest.kegg.jp/find/genes/"shiga%20toxin"
+
+### BROKEN AND
+http://rest.kegg.jp/find/genes/"shiga+toxin"
+If query is in put in quotes.
+ this result is the same as OR.
+
+### BROKEN OR
+http://rest.kegg.jp/find/genes/shiga%20toxin
+If query is not placed in quotations.
+ this result is the same as AND.
+
+## Enzyme tests
+
+http://rest.kegg.jp/find/enzyme/deoxy+1.1.1 AND works
+http://rest.kegg.jp/find/enzyme/"deoxy%201.1.1" OR does not work
+http://rest.kegg.jp/find/enzyme/"deoxy 1.1.1" OR does not work
+
+0
+http://rest.kegg.jp/find/enzyme/ec:2.4.99.1+ec:2.4.99.6 AND works but return nothing as no overlap
+http://rest.kegg.jp/find/enzyme/"ec:2.4.99.1 ec:2.4.99.6" OR doesn't work
+http://rest.kegg.jp/find/enzyme/ec:2.4.99. works
+
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/README_KEGG.md
--- a/get_data/kegg_glycan/README_KEGG.md Wed Mar 23 14:35:56 2016 -0400
+++ b/get_data/kegg_glycan/README_KEGG.md Fri May 06 08:05:48 2016 -0400
@@ -20,6 +20,9 @@
findKEGG.py -d $dbname -q $searchpattern -o $output
```
+## Known Issues
+Yes, the and/or functionality of the KEGG rest API (Release 78.0+/05-05, May 16 ) does not always function as expected. This has been resolved in the code. More details, [KEGGnotes.md ](KEGGnotes.md)
+
## Help
```
../../virtualpy/bin/activate
@@ -31,7 +34,7 @@
```
../../virtualpy/bin/activate
-python test_findKEGG.py
+nosetests test_findKEGG.py
```
# 2. getkcfKEGG,py
@@ -67,7 +70,7 @@
```
../../virtualpy/bin/activate
-python test_getkcfKEGG.py
+nosetests test_getkcfKEGG.py
```
# 3. linkKEGG.py
@@ -101,6 +104,6 @@
```
../../virtualpy/bin/activate
-python test_linkKEGG.py
+nosetests test_linkKEGG.py
```
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/db.output
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/db.output Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,83 @@
+ENTRY G10495 Glycan
+NAME 1,4-alpha-D-Glucan;
+ 1,4-alpha-Glucan;
+ Amylose;
+ Maltodextrin
+COMPOSITION (Glc)1 (*)2
+CLASS Polysaccharide
+REMARK Same as: C00718 C01935 D02329
+REACTION R06049 R06050 R06051 R06052 R06069 R06158 R06159 R06160
+ R06184 R06185 R06186 R06243
+ENZYME 2.4.1.1 2.4.1.2 2.4.1.4 2.4.1.11
+ 2.4.1.18 2.4.1.21 2.4.1.25 3.2.1.2
+ 3.2.1.33 3.2.1.54 5.4.99.15
+DBLINKS GlycomeDB: 12100
+NODE 3
+ 1 * 8 0
+ 2 Glc 0 0
+ 3 * -8 0
+EDGE 2
+ 1 2:a1 1
+ 2 3 2:4
+BRACKET 1 -4.0 2.0 -4.0 -2.0
+ 1 5.0 -2.0 5.0 2.0
+ 1 n
+///
+ENTRY G10608 Glycan
+NAME UDP-D-glucose;
+ UDP-glucose;
+ Uridine diphosphate glucose
+COMPOSITION (UDP-Glc)1
+REMARK Same as: C00029
+REACTION R06020 R06023 R06036 R06043 R06045 R06051 R06062 R06073
+ R06181 R06226 R06275 R09315 R09316
+ENZYME 2.4.1.11 2.4.1.12 2.4.1.13 2.4.1.14
+ 2.4.1.15 2.4.1.34 2.4.1.53 2.4.1.80
+ 2.4.1.183 2.4.1.188 2.4.1.-
+NODE 1
+ 1 UDP-Glc -0.3 0
+EDGE 0
+///
+ENTRY G10619 Glycan
+NAME UDP;
+ Uridine 5'-diphosphate
+COMPOSITION (UDP)1
+REMARK Same as: C00015
+REACTION R05901 R05902 R05903 R05907 R05908 R05909 R05910 R05912
+ R05915 R05916 R05925 R05926 R05927 R05928 R05929 R05930
+ R05931 R05932 R05933 R05934 R05935 R05936 R05938 R05939
+ R05941 R05946 R05948 R05952 R05953 R05956 R05960 R05962
+ R05964 R05965 R05970 R05971 R05974 R05975 R05977 R05978
+ R05983 R05985 R05986 R05987 R05989 R05991 R05992 R06006
+ R06011 R06013 R06014 R06016 R06020 R06021 R06023 R06028
+ R06029 R06033 R06036 R06043 R06045 R06051 R06055 R06062
+ R06067 R06068 R06072 R06073 R06083 R06097 R06129 R06130
+ R06145 R06167 R06168 R06169 R06172 R06173 R06174 R06181
+ R06182 R06187 R06189 R06190 R06191 R06192 R06193 R06197
+ R06198 R06210 R06214 R06226 R06234 R06275 R06276 R06277
+ R06278 R07131 R07609 R07610 R07611 R07614 R07615 R07616
+ R07617 R07619 R07621 R07622 R07623 R07624 R07625 R07628
+ R09290 R09296 R09299 R09304 R09315 R09316 R09320 R10138
+ R10139
+ENZYME 2.4.1.11 2.4.1.12 2.4.1.13 2.4.1.14
+ 2.4.1.15 2.4.1.16 2.4.1.22 2.4.1.34
+ 2.4.1.37 2.4.1.38 2.4.1.40 2.4.1.41
+ 2.4.1.43 2.4.1.45 2.4.1.47 2.4.1.53
+ 2.4.1.62 2.4.1.70 2.4.1.79 2.4.1.80
+ 2.4.1.86 2.4.1.87 2.4.1.88 2.4.1.90
+ 2.4.1.92 2.4.1.101 2.4.1.102 2.4.1.122
+ 2.4.1.123 2.4.1.133 2.4.1.134 2.4.1.135
+ 2.4.1.138 2.4.1.141 2.4.1.143 2.4.1.144
+ 2.4.1.145 2.4.1.146 2.4.1.147 2.4.1.148
+ 2.4.1.149 2.4.1.150 2.4.1.155 2.4.1.163
+ 2.4.1.164 2.4.1.165 2.4.1.167 2.4.1.174
+ 2.4.1.175 2.4.1.183 2.4.1.187 2.4.1.188
+ 2.4.1.198 2.4.1.201 2.4.1.206 2.4.1.212
+ 2.4.1.222 2.4.1.223 2.4.1.224 2.4.1.225
+ 2.4.1.226 2.4.1.227 2.4.1.228 2.4.1.244
+ 2.4.1.255 2.4.1.274 2.4.1.275 2.4.1.-
+ 2.4.2.24 2.4.2.26 2.4.2.38 2.4.2.-
+NODE 1
+ 1 UDP -0.4 0.3
+EDGE 0
+///
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/findKEGG.py
--- a/get_data/kegg_glycan/findKEGG.py Wed Mar 23 14:35:56 2016 -0400
+++ b/get_data/kegg_glycan/findKEGG.py Fri May 06 08:05:48 2016 -0400
@@ -1,6 +1,6 @@
__author__ = 'cbarnett'
__license__ = "MIT"
-__version__ = "0.3"
+__version__ = "0.4"
# http://www.kegg.jp/kegg/rest/keggapi.html
@@ -16,8 +16,18 @@
raise IOError("no db given")
if query == [] or query == "" or query is None:
raise IOError("no query given")
- query = query.replace('\n', ' ') # in case of new lines
- query = query.replace('\r', ' ') # in case of new lines
+ query = query.replace('\n', '+') # in case of new lines, assume this means the user want to "AND"
+ query = query.replace('\r', '+') # in case of new lines, assume this means the user want to "AND"
+ query = query.replace('ec: ', 'ec:') # in case of ec: spaces - e.g. issue "ec: 2.4.99.1" spaces are ok usually as allows for combination searching "2.4.99.1 2.4.99.6"
+
+ import re
+ p = re.compile(' *\+ *') # ensure no unneccesary space in an AND query, otherwise incorrect behaviour from KEGG
+ queryfix = p.subn('+', query)
+ query=queryfix[0]
+
+ if ' ' in query:
+ query='"'+query+'"' # if spaces, query must be placed in quotes, otherwise incorrect behaviour from KEGG
+ query = query.replace(' ', '%20') # previous behaviour was ignoring text after a space, rather convert to '%20' and pass on to KEGG REST service
uri = 'http://rest.kegg.jp/find/'
fulluri = uri + db + "/" + query
try:
@@ -25,7 +35,7 @@
except Exception as e:
raise urllib2.HTTPError(e.url, e.code, e.msg, e.hdrs, e.fp)
if str(response.strip()) == "":
- return None
+ return "" # change return from None to "" for easily writing to file
return response
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/findKEGG.pyc
Binary file get_data/kegg_glycan/findKEGG.pyc has changed
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/findKEGG.xml
--- a/get_data/kegg_glycan/findKEGG.xml Wed Mar 23 14:35:56 2016 -0400
+++ b/get_data/kegg_glycan/findKEGG.xml Fri May 06 08:05:48 2016 -0400
@@ -1,4 +1,4 @@
-
+
using the KEGG database
python_for_glycan_tools
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/found_entries.txt
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/found_entries.txt Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,19 @@
+ec:1.1.1.133 dTDP-4-dehydrorhamnose reductase; dTDP-4-keto-L-rhamnose reductase; dTDP-4-ketorhamnose reductase; TDP-4-keto-rhamnose reductase; thymidine diphospho-4-ketorhamnose reductase; dTDP-6-deoxy-L-mannose:NADP+ 4-oxidoreductase; dTDP-6-deoxy-beta-L-mannose:NADP+ 4-oxidoreductase
+ec:1.1.1.340 1-deoxy-11beta-hydroxypentalenate dehydrogenase; 1-deoxy-11beta-hydroxypentalenic acid dehydrogenase; ptlF (gene name); penF (gene name)
+ec:1.14.11.35 1-deoxypentalenic acid 11beta-hydroxylase; ptlH (gene name); sav2991 (gene name); pntH (gene name)
+ec:2.3.1.202 UDP-4-amino-4,6-dideoxy-N-acetyl-beta-L-altrosamine N-acetyltransferase; PseH
+ec:2.4.1.27 DNA beta-glucosyltransferase; T4-HMC-beta-glucosyl transferase; T4-beta-glucosyl transferase; T4 phage beta-glucosyltransferase; UDP glucose-DNA beta-glucosyltransferase; uridine diphosphoglucose-deoxyribonucleate beta-glucosyltransferase
+ec:2.4.1.28 glucosyl-DNA beta-glucosyltransferase; T6-glucosyl-HMC-beta-glucosyl transferase; T6-beta-glucosyl transferase; uridine diphosphoglucose-glucosyldeoxyribonucleate beta-glucosyltransferase
+ec:2.4.1.335 dolichyl N-acetyl-alpha-D-glucosaminyl phosphate 3-beta-D-2,3-diacetamido-2,3-dideoxy-beta-D-glucuronosyltransferase; AglC; UDP-Glc-2,3-diNAcA glycosyltransferase
+ec:2.6.1.87 UDP-4-amino-4-deoxy-L-arabinose aminotransferase; UDP-(beta-L-threo-pentapyranosyl-4''-ulose diphosphate) aminotransferase; UDP-4-amino-4-deoxy-L-arabinose---oxoglutarate aminotransferase; UDP-Ara4O aminotransferase; UDP-L-Ara4N transaminase
+ec:2.6.1.92 UDP-4-amino-4,6-dideoxy-N-acetyl-beta-L-altrosamine transaminase; PseC
+ec:2.7.7.7 DNA-directed DNA polymerase; DNA polymerase I; DNA polymerase II; DNA polymerase III; DNA polymerase alpha; DNA polymerase beta; DNA polymerase gamma; DNA nucleotidyltransferase (DNA-directed); DNA nucleotidyltransferase (DNA-directed); deoxyribonucleate nucleotidyltransferase; deoxynucleate polymerase; deoxyribonucleic acid duplicase; deoxyribonucleic acid polymerase; deoxyribonucleic duplicase; deoxyribonucleic polymerase; deoxyribonucleic polymerase I; DNA duplicase; DNA nucleotidyltransferase; DNA polymerase; DNA replicase; DNA-dependent DNA polymerase; duplicase; Klenow fragment; sequenase; Taq DNA polymerase; Taq Pol I; Tca DNA polymerase
+ec:3.1.6.9 chondro-4-sulfatase; chondroitin-4-sulfatase; 4-deoxy-beta-D-gluc-4-enuronosyl-(1,3)-N-acetyl-D-galactosamine-4-sulfate 4-sulfohydrolase
+ec:3.1.6.10 chondro-6-sulfatase; 4-deoxy-beta-D-gluc-4-enuronosyl-(1,3)-N-acetyl-D-galactosamine-6-sulfate 6-sulfohydrolase
+ec:3.2.1.52 beta-N-acetylhexosaminidase; hexosaminidase; beta-acetylaminodeoxyhexosidase; N-acetyl-beta-D-hexosaminidase; N-acetyl-beta-hexosaminidase; beta-hexosaminidase; beta-acetylhexosaminidinase; beta-D-N-acetylhexosaminidase; beta-N-acetyl-D-hexosaminidase; beta-N-acetylglucosaminidase; hexosaminidase A; N-acetylhexosaminidase; beta-D-hexosaminidase
+ec:3.2.1.92 peptidoglycan beta-N-acetylmuramidase; exo-beta-N-acetylmuramidase; exo-beta-acetylmuramidase; beta-2-acetamido-3-O-(D-1-carboxyethyl)-2-deoxy-D-glucoside acetamidodeoxyglucohydrolase
+ec:3.2.1.96 mannosyl-glycoprotein endo-beta-N-acetylglucosaminidase; N,N'-diacetylchitobiosyl beta-N-acetylglucosaminidase; endo-beta-N-acetylglucosaminidase; mannosyl-glycoprotein endo-beta-N-acetylglucosamidase; di-N-acetylchitobiosyl beta-N-acetylglucosaminidase; endo-beta-acetylglucosaminidase; endo-beta-(1->4)-N-acetylglucosaminidase; mannosyl-glycoprotein 1,4-N-acetamidodeoxy-beta-D-glycohydrolase; endoglycosidase S; endo-N-acetyl-beta-D-glucosaminidase; endo-N-acetyl-beta-glucosaminidase; endo-beta-N-acetylglucosaminidase D; endo-beta-N-acetylglucosaminidase F; endo-beta-N-acetylglucosaminidase H; endo-beta-N-acetylglucosaminidase L; glycopeptide-D-mannosyl-4-N-(N-acetyl-D-glucosaminyl)2-asparagine 1,4-N-acetyl-beta-glucosaminohydrolase; endoglycosidase H
+ec:3.5.1.105 chitin disaccharide deacetylase; chitobiose amidohydolase; COD; chitin oligosaccharide deacetylase; chitin oligosaccharide amidohydolase; 2-(acetylamino)-4-O-[2-(acetylamino)-2-deoxy-beta-D-glucopyranosyl]-2-deoxy-D-glucopyranose acetylhydrolase
+ec:3.6.1.57 UDP-2,4-diacetamido-2,4,6-trideoxy-beta-L-altropyranose hydrolase; PseG; UDP-6-deoxy-AltdiNAc hydrolase; Cj1312; UDP-2,4-bis(acetamido)-2,4,6-trideoxy-beta-L-altropyranose hydrolase
+ec:4.1.2.20 2-dehydro-3-deoxyglucarate aldolase; 2-keto-3-deoxyglucarate aldolase; alpha-keto-beta-deoxy-D-glucarate aldolase; 2-dehydro-3-deoxy-D-glucarate tartronate-semialdehyde-lyase; 2-dehydro-3-deoxy-D-glucarate tartronate-semialdehyde-lyase (pyruvate-forming)
+ec:4.2.1.115 UDP-N-acetylglucosamine 4,6-dehydratase (configuration-inverting); FlaA1; UDP-N-acetylglucosamine 5-inverting 4,6-dehydratase; PseB; UDP-N-acetylglucosamine hydro-lyase (inverting; UDP-2-acetamido-2,6-dideoxy-beta-L-arabino-hex-4-ulose-forming)
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/getkcfKEGG.pyc
Binary file get_data/kegg_glycan/getkcfKEGG.pyc has changed
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/kcf.output
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/kcf.output Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,22 @@
+ENTRY G10495 Glycan
+NODE 3
+ 1 * 8 0
+ 2 Glc 0 0
+ 3 * -8 0
+EDGE 2
+ 1 2:a1 1
+ 2 3 2:4
+BRACKET 1 -4.0 2.0 -4.0 -2.0
+ 1 5.0 -2.0 5.0 2.0
+ 1 n
+///
+ENTRY G10608 Glycan
+NODE 1
+ 1 UDP-Glc -0.3 0
+EDGE 0
+///
+ENTRY G10619 Glycan
+NODE 1
+ 1 UDP -0.4 0.3
+EDGE 0
+///
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/linkKEGG.py
--- a/get_data/kegg_glycan/linkKEGG.py Wed Mar 23 14:35:56 2016 -0400
+++ b/get_data/kegg_glycan/linkKEGG.py Fri May 06 08:05:48 2016 -0400
@@ -1,6 +1,6 @@
__author__ = 'cbarnett'
__license__ = "MIT"
-__version = "0.3"
+__version = "0.3.1"
# http://www.kegg.jp/kegg/rest/keggapi.html
@@ -13,6 +13,15 @@
import urllib2
uri = 'http://rest.kegg.jp/link/'
+
+ import re
+ p = re.compile(' *\+ *') # ensure no unneccessary space in an AND query
+ sdbfix = p.subn('+', sourcedb)
+ sourcedb=sdbfix[0]
+
+ if ' ' in sourcedb:
+ sourcedb = sourcedb.replace(' ', '%20') # previous behaviour was ignoring text after a space, rather convert to '%20' and pass on to KEGG REST service
+
fulluri = uri + targetdb + "/" + sourcedb
try:
response = urllib2.urlopen(fulluri).read()
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/linkKEGG.pyc
Binary file get_data/kegg_glycan/linkKEGG.pyc has changed
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/linkKEGG.xml
--- a/get_data/kegg_glycan/linkKEGG.xml Wed Mar 23 14:35:56 2016 -0400
+++ b/get_data/kegg_glycan/linkKEGG.xml Fri May 06 08:05:48 2016 -0400
@@ -1,4 +1,4 @@
-
+
searches KEGG for related database entries
python_for_glycan_tools
@@ -34,7 +34,7 @@
-
+
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/linked_entries.txt
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/plus
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/plus Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,21 @@
+ec:1.1.1.125 2-deoxy-D-gluconate 3-dehydrogenase; 2-deoxygluconate dehydrogenase
+ec:1.1.1.126 2-dehydro-3-deoxy-D-gluconate 6-dehydrogenase; 2-keto-3-deoxy-D-gluconate dehydrogenase; 2-keto-3-deoxygluconate dehydrogenase
+ec:1.1.1.127 2-dehydro-3-deoxy-D-gluconate 5-dehydrogenase; 2-keto-3-deoxygluconate 5-dehydrogenase; 2-keto-3-deoxy-D-gluconate dehydrogenase; 2-keto-3-deoxygluconate (nicotinamide adenine dinucleotide (phosphate)) dehydrogenase; 2-keto-3-deoxy-D-gluconate (3-deoxy-D-glycero-2,5-hexodiulosonic acid) dehydrogenase
+ec:1.1.1.133 dTDP-4-dehydrorhamnose reductase; dTDP-4-keto-L-rhamnose reductase; dTDP-4-ketorhamnose reductase; TDP-4-keto-rhamnose reductase; thymidine diphospho-4-ketorhamnose reductase; dTDP-6-deoxy-L-mannose:NADP+ 4-oxidoreductase; dTDP-6-deoxy-beta-L-mannose:NADP+ 4-oxidoreductase
+ec:1.1.1.134 dTDP-6-deoxy-L-talose 4-dehydrogenase (NADP+); thymidine diphospho-6-deoxy-L-talose dehydrogenase; TDP-6-deoxy-L-talose dehydrogenase; dTDP-6-deoxy-L-talose dehydrogenase (4-reductase); dTDP-6-deoxy-L-talose:NADP+ 4-oxidoreductase
+ec:1.1.1.135 GDP-6-deoxy-D-talose 4-dehydrogenase; guanosine diphospho-6-deoxy-D-talose dehydrogenase; GDP-6-deoxy-D-talose:NAD(P)+ 4-oxidoreductase
+ec:1.1.1.136 UDP-N-acetylglucosamine 6-dehydrogenase; uridine diphosphoacetylglucosamine dehydrogenase; UDP-acetylglucosamine dehydrogenase; UDP-2-acetamido-2-deoxy-D-glucose:NAD oxidoreductase; UDP-GlcNAc dehydrogenase; WbpA; WbpO
+ec:1.1.1.187 GDP-4-dehydro-D-rhamnose reductase; GDP-4-keto-6-deoxy-D-mannose reductase; GDP-4-keto-D-rhamnose reductase; guanosine diphosphate-4-keto-D-rhamnose reductase; GDP-6-deoxy-D-mannose:NAD(P)+ 4-oxidoreductase; GDP-6-deoxy-alpha-D-mannose:NAD(P)+ 4-oxidoreductase
+ec:1.1.1.235 8-oxocoformycin reductase; 8-ketodeoxycoformycin reductase
+ec:1.1.1.266 dTDP-4-dehydro-6-deoxyglucose reductase; dTDP-4-keto-6-deoxyglucose reductase; dTDP-D-fucose:NADP+ oxidoreductase; Fcf1; dTDP-6-deoxy-D-xylo-hex-4-ulopyranose reductase
+ec:1.1.1.267 1-deoxy-D-xylulose-5-phosphate reductoisomerase; DXP-reductoisomerase; 1-deoxy-D-xylulose-5-phosphate isomeroreductase; 2-C-methyl-D-erythritol 4-phosphate (MEP) synthase
+ec:1.1.1.271 GDP-L-fucose synthase; GDP-4-keto-6-deoxy-D-mannose-3,5-epimerase-4-reductase; GDP-L-fucose:NADP+ 4-oxidoreductase (3,5-epimerizing)
+ec:1.1.1.281 GDP-4-dehydro-6-deoxy-D-mannose reductase; GDP-4-keto-6-deoxy-D-mannose reductase [ambiguous]; GDP-6-deoxy-D-lyxo-4-hexulose reductase; Rmd; GDP-6-deoxy-D-mannose:NAD(P)+ 4-oxidoreductase (D-rhamnose-forming); GDP-6-deoxy-alpha-D-mannose:NAD(P)+ 4-oxidoreductase (D-rhamnose-forming)
+ec:1.1.1.329 2-deoxy-scyllo-inosamine dehydrogenase; neoA (gene name); kanK (gene name, ambiguous); kanE (gene name, ambiguous)
+ec:1.1.1.335 UDP-N-acetyl-2-amino-2-deoxyglucuronate dehydrogenase; WlbA; WbpB
+ec:1.1.1.339 dTDP-6-deoxy-L-talose 4-dehydrogenase (NAD+); tll (gene name)
+ec:1.1.1.340 1-deoxy-11beta-hydroxypentalenate dehydrogenase; 1-deoxy-11beta-hydroxypentalenic acid dehydrogenase; ptlF (gene name); penF (gene name)
+ec:1.1.1.344 dTDP-6-deoxy-L-talose 4-dehydrogenase [NAD(P)+]; tal (gene name)
+ec:1.1.1.364 dTDP-4-dehydro-6-deoxy-alpha-D-gulose 4-ketoreductase; dTDP-4-dehydro-6-deoxygulose reductase; tylD (gene name); gerKI (gene name); chmD (gene name); mydI (gene name)
+ec:1.1.1.384 dTDP-3,4-didehydro-2,6-dideoxy-alpha-D-glucose 3-reductase; KijD10; dTDP-4-keto-2,6-dideoxy-D-glucose 3-oxidoreductase; dTDP-4-dehydro-2,6-dideoxy-alpha-D-glucose 3-oxidoreductase
+ec:1.1.1.389 2-dehydro-3-deoxy-L-galactonate 5-dehydrogenase
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/test-data/shiga-toxin-example-AND
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/test-data/shiga-toxin-example-AND Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,103 @@
+ece:Z1464 stx2A; shiga-like toxin II A subunit encoded by bacteriophage BP-933W; K11006 shiga toxin subunit A
+ece:Z1465 stx2B; shiga-like toxin II B subunit encoded by bacteriophage BP-933W; K11007 shiga toxin subunit B
+ece:Z3343 stx1B; shiga-like toxin 1 subunit B encoded within prophage CP-933V; K11007 shiga toxin subunit B
+ece:Z3344 stx1A; shiga-like toxin 1 subunit A encoded within prophage CP-933V; K11006 shiga toxin subunit A
+ecs:ECs1205 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+ecs:ECs1206 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+ecs:ECs2973 Shiga toxin I subunit B; K11007 shiga toxin subunit B
+ecs:ECs2974 Shiga toxin I subunit A; K11006 shiga toxin subunit A
+ecf:ECH74115_2905 shigatoxin 2, subunit B; K11007 shiga toxin subunit B
+ecf:ECH74115_2906 shiga toxin subunit A; K11006 shiga toxin subunit A
+ecf:ECH74115_3532 shiga toxin 2 B subunit; K11007 shiga toxin subunit B
+ecf:ECH74115_3533 shiga toxin subunit A; K11006 shiga toxin subunit A
+etw:ECSP_2722 stx2cB; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+etw:ECSP_2723 stx2A1; Shiga-like toxin II subunit A precursor; K11006 shiga toxin subunit A
+etw:ECSP_3252 stx2B; shiga toxin II subunit B; K11007 shiga toxin subunit B
+etw:ECSP_3253 stx2A2; shiga toxin II subunit A; K11006 shiga toxin subunit A
+elx:CDCO157_1154 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+elx:CDCO157_1155 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+elx:CDCO157_2738 Shiga toxin I subunit B precursor; K11007 shiga toxin subunit B
+elx:CDCO157_2739 Shiga toxin I subunit A precursor; K11006 shiga toxin subunit A
+eoj:ECO26_1599 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoj:ECO26_1600 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eoi:ECO111_2429 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eoi:ECO111_2430 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+eoi:ECO111_3361 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoi:ECO111_3362 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eoh:ECO103_2844 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eoh:ECO103_2845 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+eoh:ECO103_5197 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoh:ECO103_5198 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eso:O3O_09170 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eso:O3O_09165 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+esm:O3M_16105 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+esm:O3M_16110 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+esl:O3K_16130 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+esl:O3K_16135 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+ecoo:ECRM13514_3144 stx2B; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+ecoo:ECRM13514_3145 stx2A; Shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+ecoh:ECRM13516_2315 stx2B; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+ecoh:ECRM13516_2316 stx2A; Shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+stm:STM0284 shiga-like toxin A subunit
+seo:STM14_0332 putative Shiga-like toxin A subunit
+sev:STMMW_02891 putative shiga-like toxin A subunit
+sem:STMDT12_C02790 putative Shiga-like toxin A subunit
+sej:STMUK_0287 putative Shiga-like toxin A subunit
+setu:STU288_13325 putative shiga-like toxin A subunit
+setc:CFSAN001921_15985 Shiga toxin A subunit
+send:DT104_02831 sciR; shiga-like toxin a subunit, type VI secretion system SciR protein flags: precursor
+seni:CY43_01420 Shiga toxin A subunit
+seen:SE451236_07435 Shiga toxin A subunit
+spt:SPA2480 putative shiga-like toxin A subunit
+sek:SSPA2315 shiga-like toxin A subunit
+sei:SPC_0292 putative shiga-like toxin A subunit
+sec:SCH_0277 putative shiga-like toxin A subunit
+shb:SU5_0926 SciR Shiga-like toxin A subunit
+senh:CFSAN002069_07720 Shiga toxin A subunit
+seeh:SEEH1578_10515 putative shiga-like toxin A subunit
+seec:CFSAN002050_07960 Shiga toxin A subunit
+sene:IA1_01505 Shiga toxin A subunit
+senc:SEET0819_08350 Shiga toxin A subunit
+sdy:SDY_1389 stxA; Shiga toxin subunit A; K11006 shiga toxin subunit A
+sdy:SDY_1390 stxB; Shiga toxin subunit B; K11007 shiga toxin subunit B
+sdz:Asd1617_01826 Shiga toxin subunit A (EC:3.2.2.22); K11006 shiga toxin subunit A
+ebi:EbC_13780 Putative shiga-like toxin A subunit
+psos:POS17_2009 P9; putative Shiga-like toxin alpha subunit
+ssab:SSABA_v1c03870 hypothetical protein; K11006 shiga toxin subunit A
+ssab:SSABA_v1c09020 hypothetical protein; K11006 shiga toxin subunit A
+calt:Cal6303_2132 ribosome-inactivating protein; K11006 shiga toxin subunit A
+vg:22277199 PI30_gp34; Shigella phage POCJ13; shiga toxin A-chain precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:22277200 PI30_gp35; Shigella phage POCJ13; shiga toxin I subunit B precursor; K11007 shiga toxin subunit B
+vg:1262301 APSE-1_07; Endosymbiont phage APSE-1; P7; K11007 shiga toxin subunit B
+vg:1481747 stx1A, Stx1_p142; Escherichia Stx1 converting phage; Shiga toxin1 subunit A; K11006 shiga toxin subunit A
+vg:1481748 stx1B, Stx1_p143; Escherichia Stx1 converting phage; Shiga toxin1 subunit B; K11007 shiga toxin subunit B
+vg:1481767 stx2A, Stx2II_p143; Stx2 converting phage II; Shiga toxin2 subunit A; K11006 shiga toxin subunit A
+vg:1481768 stx2B, Stx2II_p144; Stx2 converting phage II; Shiga toxin2 subunit B; K11007 shiga toxin subunit B
+vg:26228052 AS349_gp84; Escherichia phage PA2; shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:26228100 AS349_gp83; Escherichia phage PA2; shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+vg:26516496 stx2A, AU085_gp42; Enterobacteria phage VT2phi_272; Shiga toxin 2 A subunit; K11006 shiga toxin subunit A
+vg:26516448 stx2B, AU085_gp41; Enterobacteria phage VT2phi_272; Shiga toxin 2 B subunit; K11007 shiga toxin subunit B
+vg:26516283 stxB, phi191_00011, AU083_gp11; Escherichia phage phi191; shiga toxin subunit B precursor; K11007 shiga toxin subunit B
+vg:26516284 stxA, phi191_00012, AU083_gp12; Escherichia phage phi191; shiga toxin subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:26519429 AU154_gp39; Shigella phage Ss-VASD; Stx1 A subunit; K11006 shiga toxin subunit A
+vg:26519430 AU154_gp40; Shigella phage Ss-VASD; Stx1 B subunit; K11007 shiga toxin subunit B
+vg:1261950 stxA2, L0103, sltIIA, stx2A, 933Wp40; Enterobacteria phage 933W; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:1262010 stxB2, L0104, sltIIB, stx2B, 933Wp41; Enterobacteria phage 933W; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:2641645 stxA1, phi4795p38, PBV4795_ORF40; Enterobacteria phage BP-4795; Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+vg:2641657 stxB1, phi4795p39, PBV4795_ORF41; Enterobacteria phage BP-4795; Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+vg:929695 stxA2e, P27p25; Enterobacteria phage phiP27; shiga toxin 2e A-subunit; K11006 shiga toxin subunit A
+vg:929727 stxB2e, P27p26; Enterobacteria phage phiP27; shiga toxin 2e B-subunit; K11007 shiga toxin subunit B
+vg:4397483 stx2A, Stx2-86_gp01; Stx2-converting phage 86; shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:4397484 stx2B, Stx2-86_gp02; Stx2-converting phage 86; shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:6159405 stx2A, pMIN27_41; Enterobacteria phage Min27; ORF44; K11006 shiga toxin subunit A
+vg:6159351 stx2B, pMIN27_42; Enterobacteria phage Min27; ORF45; K11007 shiga toxin subunit B
+vg:6973138 Stx2-1717_gp41; Stx2-converting phage 1717; verocytotoxin 2 variant 2c subunit A; K11006 shiga toxin subunit A
+vg:6972909 stx2cB, Stx2-1717_gp42; Stx2-converting phage 1717; verocytotoxin 2 variant 2c subunit B; K11007 shiga toxin subunit B
+vg:6973079 YYZ_gp39; Enterobacteria phage YYZ-2008; Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+vg:6973080 YYZ_gp40; Enterobacteria phage YYZ-2008; Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+vg:13828571 stx2A, D300_gp43; Escherichia phage P13374; shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:13828535 stx2B, D300_gp42; Escherichia phage P13374; shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:14005228 F366_gp36; Escherichia phage TL-2011c; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:14005229 F366_gp37; Escherichia phage TL-2011c; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:1262249 stx2A, VT2-Sap42; Enterobacteria phage VT2-Sakai; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:1262250 stx2B, VT2-Sap43; Enterobacteria phage VT2-Sakai; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/test-data/shiga-toxin-example-AND-broken
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/test-data/shiga-toxin-example-AND-broken Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,88 @@
+ece:Z1464 stx2A; shiga-like toxin II A subunit encoded by bacteriophage BP-933W; K11006 shiga toxin subunit A
+ece:Z1465 stx2B; shiga-like toxin II B subunit encoded by bacteriophage BP-933W; K11007 shiga toxin subunit B
+ece:Z3343 stx1B; shiga-like toxin 1 subunit B encoded within prophage CP-933V; K11007 shiga toxin subunit B
+ece:Z3344 stx1A; shiga-like toxin 1 subunit A encoded within prophage CP-933V; K11006 shiga toxin subunit A
+ecs:ECs1205 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+ecs:ECs1206 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+ecs:ECs2973 Shiga toxin I subunit B; K11007 shiga toxin subunit B
+ecs:ECs2974 Shiga toxin I subunit A; K11006 shiga toxin subunit A
+ecf:ECH74115_2905 shigatoxin 2, subunit B; K11007 shiga toxin subunit B
+ecf:ECH74115_2906 shiga toxin subunit A; K11006 shiga toxin subunit A
+ecf:ECH74115_3532 shiga toxin 2 B subunit; K11007 shiga toxin subunit B
+ecf:ECH74115_3533 shiga toxin subunit A; K11006 shiga toxin subunit A
+etw:ECSP_2722 stx2cB; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+etw:ECSP_2723 stx2A1; Shiga-like toxin II subunit A precursor; K11006 shiga toxin subunit A
+etw:ECSP_3252 stx2B; shiga toxin II subunit B; K11007 shiga toxin subunit B
+etw:ECSP_3253 stx2A2; shiga toxin II subunit A; K11006 shiga toxin subunit A
+elx:CDCO157_1154 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+elx:CDCO157_1155 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+elx:CDCO157_2738 Shiga toxin I subunit B precursor; K11007 shiga toxin subunit B
+elx:CDCO157_2739 Shiga toxin I subunit A precursor; K11006 shiga toxin subunit A
+eoj:ECO26_1599 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoj:ECO26_1600 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eoi:ECO111_2429 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eoi:ECO111_2430 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+eoi:ECO111_3361 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoi:ECO111_3362 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eoh:ECO103_2844 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eoh:ECO103_2845 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+eoh:ECO103_5197 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoh:ECO103_5198 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eso:O3O_09170 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eso:O3O_09165 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+esm:O3M_16105 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+esm:O3M_16110 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+esl:O3K_16130 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+esl:O3K_16135 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+ecoo:ECRM13514_3144 stx2B; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+ecoo:ECRM13514_3145 stx2A; Shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+ecoh:ECRM13516_2315 stx2B; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+ecoh:ECRM13516_2316 stx2A; Shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+setc:CFSAN001921_15985 Shiga toxin A subunit
+seni:CY43_01420 Shiga toxin A subunit
+seen:SE451236_07435 Shiga toxin A subunit
+senh:CFSAN002069_07720 Shiga toxin A subunit
+seec:CFSAN002050_07960 Shiga toxin A subunit
+sene:IA1_01505 Shiga toxin A subunit
+senc:SEET0819_08350 Shiga toxin A subunit
+sdy:SDY_1389 stxA; Shiga toxin subunit A; K11006 shiga toxin subunit A
+sdy:SDY_1390 stxB; Shiga toxin subunit B; K11007 shiga toxin subunit B
+sdz:Asd1617_01826 Shiga toxin subunit A (EC:3.2.2.22); K11006 shiga toxin subunit A
+ssab:SSABA_v1c03870 hypothetical protein; K11006 shiga toxin subunit A
+ssab:SSABA_v1c09020 hypothetical protein; K11006 shiga toxin subunit A
+calt:Cal6303_2132 ribosome-inactivating protein; K11006 shiga toxin subunit A
+vg:22277199 PI30_gp34; Shigella phage POCJ13; shiga toxin A-chain precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:22277200 PI30_gp35; Shigella phage POCJ13; shiga toxin I subunit B precursor; K11007 shiga toxin subunit B
+vg:1262301 APSE-1_07; Endosymbiont phage APSE-1; P7; K11007 shiga toxin subunit B
+vg:1481747 stx1A, Stx1_p142; Escherichia Stx1 converting phage; Shiga toxin1 subunit A; K11006 shiga toxin subunit A
+vg:1481748 stx1B, Stx1_p143; Escherichia Stx1 converting phage; Shiga toxin1 subunit B; K11007 shiga toxin subunit B
+vg:1481767 stx2A, Stx2II_p143; Stx2 converting phage II; Shiga toxin2 subunit A; K11006 shiga toxin subunit A
+vg:1481768 stx2B, Stx2II_p144; Stx2 converting phage II; Shiga toxin2 subunit B; K11007 shiga toxin subunit B
+vg:26228052 AS349_gp84; Escherichia phage PA2; shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:26228100 AS349_gp83; Escherichia phage PA2; shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+vg:26516496 stx2A, AU085_gp42; Enterobacteria phage VT2phi_272; Shiga toxin 2 A subunit; K11006 shiga toxin subunit A
+vg:26516448 stx2B, AU085_gp41; Enterobacteria phage VT2phi_272; Shiga toxin 2 B subunit; K11007 shiga toxin subunit B
+vg:26516283 stxB, phi191_00011, AU083_gp11; Escherichia phage phi191; shiga toxin subunit B precursor; K11007 shiga toxin subunit B
+vg:26516284 stxA, phi191_00012, AU083_gp12; Escherichia phage phi191; shiga toxin subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:26519429 AU154_gp39; Shigella phage Ss-VASD; Stx1 A subunit; K11006 shiga toxin subunit A
+vg:26519430 AU154_gp40; Shigella phage Ss-VASD; Stx1 B subunit; K11007 shiga toxin subunit B
+vg:1261950 stxA2, L0103, sltIIA, stx2A, 933Wp40; Enterobacteria phage 933W; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:1262010 stxB2, L0104, sltIIB, stx2B, 933Wp41; Enterobacteria phage 933W; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:2641645 stxA1, phi4795p38, PBV4795_ORF40; Enterobacteria phage BP-4795; Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+vg:2641657 stxB1, phi4795p39, PBV4795_ORF41; Enterobacteria phage BP-4795; Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+vg:929695 stxA2e, P27p25; Enterobacteria phage phiP27; shiga toxin 2e A-subunit; K11006 shiga toxin subunit A
+vg:929727 stxB2e, P27p26; Enterobacteria phage phiP27; shiga toxin 2e B-subunit; K11007 shiga toxin subunit B
+vg:4397483 stx2A, Stx2-86_gp01; Stx2-converting phage 86; shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:4397484 stx2B, Stx2-86_gp02; Stx2-converting phage 86; shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:6159405 stx2A, pMIN27_41; Enterobacteria phage Min27; ORF44; K11006 shiga toxin subunit A
+vg:6159351 stx2B, pMIN27_42; Enterobacteria phage Min27; ORF45; K11007 shiga toxin subunit B
+vg:6973138 Stx2-1717_gp41; Stx2-converting phage 1717; verocytotoxin 2 variant 2c subunit A; K11006 shiga toxin subunit A
+vg:6972909 stx2cB, Stx2-1717_gp42; Stx2-converting phage 1717; verocytotoxin 2 variant 2c subunit B; K11007 shiga toxin subunit B
+vg:6973079 YYZ_gp39; Enterobacteria phage YYZ-2008; Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+vg:6973080 YYZ_gp40; Enterobacteria phage YYZ-2008; Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+vg:13828571 stx2A, D300_gp43; Escherichia phage P13374; shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:13828535 stx2B, D300_gp42; Escherichia phage P13374; shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:14005228 F366_gp36; Escherichia phage TL-2011c; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:14005229 F366_gp37; Escherichia phage TL-2011c; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:1262249 stx2A, VT2-Sap42; Enterobacteria phage VT2-Sakai; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:1262250 stx2B, VT2-Sap43; Enterobacteria phage VT2-Sakai; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/test-data/shiga-toxin-example-OR
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/test-data/shiga-toxin-example-OR Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,88 @@
+ece:Z1464 stx2A; shiga-like toxin II A subunit encoded by bacteriophage BP-933W; K11006 shiga toxin subunit A
+ece:Z1465 stx2B; shiga-like toxin II B subunit encoded by bacteriophage BP-933W; K11007 shiga toxin subunit B
+ece:Z3343 stx1B; shiga-like toxin 1 subunit B encoded within prophage CP-933V; K11007 shiga toxin subunit B
+ece:Z3344 stx1A; shiga-like toxin 1 subunit A encoded within prophage CP-933V; K11006 shiga toxin subunit A
+ecs:ECs1205 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+ecs:ECs1206 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+ecs:ECs2973 Shiga toxin I subunit B; K11007 shiga toxin subunit B
+ecs:ECs2974 Shiga toxin I subunit A; K11006 shiga toxin subunit A
+ecf:ECH74115_2905 shigatoxin 2, subunit B; K11007 shiga toxin subunit B
+ecf:ECH74115_2906 shiga toxin subunit A; K11006 shiga toxin subunit A
+ecf:ECH74115_3532 shiga toxin 2 B subunit; K11007 shiga toxin subunit B
+ecf:ECH74115_3533 shiga toxin subunit A; K11006 shiga toxin subunit A
+etw:ECSP_2722 stx2cB; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+etw:ECSP_2723 stx2A1; Shiga-like toxin II subunit A precursor; K11006 shiga toxin subunit A
+etw:ECSP_3252 stx2B; shiga toxin II subunit B; K11007 shiga toxin subunit B
+etw:ECSP_3253 stx2A2; shiga toxin II subunit A; K11006 shiga toxin subunit A
+elx:CDCO157_1154 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+elx:CDCO157_1155 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+elx:CDCO157_2738 Shiga toxin I subunit B precursor; K11007 shiga toxin subunit B
+elx:CDCO157_2739 Shiga toxin I subunit A precursor; K11006 shiga toxin subunit A
+eoj:ECO26_1599 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoj:ECO26_1600 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eoi:ECO111_2429 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eoi:ECO111_2430 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+eoi:ECO111_3361 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoi:ECO111_3362 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eoh:ECO103_2844 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eoh:ECO103_2845 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+eoh:ECO103_5197 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoh:ECO103_5198 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eso:O3O_09170 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eso:O3O_09165 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+esm:O3M_16105 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+esm:O3M_16110 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+esl:O3K_16130 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+esl:O3K_16135 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+ecoo:ECRM13514_3144 stx2B; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+ecoo:ECRM13514_3145 stx2A; Shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+ecoh:ECRM13516_2315 stx2B; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+ecoh:ECRM13516_2316 stx2A; Shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+setc:CFSAN001921_15985 Shiga toxin A subunit
+seni:CY43_01420 Shiga toxin A subunit
+seen:SE451236_07435 Shiga toxin A subunit
+senh:CFSAN002069_07720 Shiga toxin A subunit
+seec:CFSAN002050_07960 Shiga toxin A subunit
+sene:IA1_01505 Shiga toxin A subunit
+senc:SEET0819_08350 Shiga toxin A subunit
+sdy:SDY_1389 stxA; Shiga toxin subunit A; K11006 shiga toxin subunit A
+sdy:SDY_1390 stxB; Shiga toxin subunit B; K11007 shiga toxin subunit B
+sdz:Asd1617_01826 Shiga toxin subunit A (EC:3.2.2.22); K11006 shiga toxin subunit A
+ssab:SSABA_v1c03870 hypothetical protein; K11006 shiga toxin subunit A
+ssab:SSABA_v1c09020 hypothetical protein; K11006 shiga toxin subunit A
+calt:Cal6303_2132 ribosome-inactivating protein; K11006 shiga toxin subunit A
+vg:22277199 PI30_gp34; Shigella phage POCJ13; shiga toxin A-chain precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:22277200 PI30_gp35; Shigella phage POCJ13; shiga toxin I subunit B precursor; K11007 shiga toxin subunit B
+vg:1262301 APSE-1_07; Endosymbiont phage APSE-1; P7; K11007 shiga toxin subunit B
+vg:1481747 stx1A, Stx1_p142; Escherichia Stx1 converting phage; Shiga toxin1 subunit A; K11006 shiga toxin subunit A
+vg:1481748 stx1B, Stx1_p143; Escherichia Stx1 converting phage; Shiga toxin1 subunit B; K11007 shiga toxin subunit B
+vg:1481767 stx2A, Stx2II_p143; Stx2 converting phage II; Shiga toxin2 subunit A; K11006 shiga toxin subunit A
+vg:1481768 stx2B, Stx2II_p144; Stx2 converting phage II; Shiga toxin2 subunit B; K11007 shiga toxin subunit B
+vg:26228052 AS349_gp84; Escherichia phage PA2; shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:26228100 AS349_gp83; Escherichia phage PA2; shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+vg:26516496 stx2A, AU085_gp42; Enterobacteria phage VT2phi_272; Shiga toxin 2 A subunit; K11006 shiga toxin subunit A
+vg:26516448 stx2B, AU085_gp41; Enterobacteria phage VT2phi_272; Shiga toxin 2 B subunit; K11007 shiga toxin subunit B
+vg:26516283 stxB, phi191_00011, AU083_gp11; Escherichia phage phi191; shiga toxin subunit B precursor; K11007 shiga toxin subunit B
+vg:26516284 stxA, phi191_00012, AU083_gp12; Escherichia phage phi191; shiga toxin subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:26519429 AU154_gp39; Shigella phage Ss-VASD; Stx1 A subunit; K11006 shiga toxin subunit A
+vg:26519430 AU154_gp40; Shigella phage Ss-VASD; Stx1 B subunit; K11007 shiga toxin subunit B
+vg:1261950 stxA2, L0103, sltIIA, stx2A, 933Wp40; Enterobacteria phage 933W; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:1262010 stxB2, L0104, sltIIB, stx2B, 933Wp41; Enterobacteria phage 933W; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:2641645 stxA1, phi4795p38, PBV4795_ORF40; Enterobacteria phage BP-4795; Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+vg:2641657 stxB1, phi4795p39, PBV4795_ORF41; Enterobacteria phage BP-4795; Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+vg:929695 stxA2e, P27p25; Enterobacteria phage phiP27; shiga toxin 2e A-subunit; K11006 shiga toxin subunit A
+vg:929727 stxB2e, P27p26; Enterobacteria phage phiP27; shiga toxin 2e B-subunit; K11007 shiga toxin subunit B
+vg:4397483 stx2A, Stx2-86_gp01; Stx2-converting phage 86; shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:4397484 stx2B, Stx2-86_gp02; Stx2-converting phage 86; shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:6159405 stx2A, pMIN27_41; Enterobacteria phage Min27; ORF44; K11006 shiga toxin subunit A
+vg:6159351 stx2B, pMIN27_42; Enterobacteria phage Min27; ORF45; K11007 shiga toxin subunit B
+vg:6973138 Stx2-1717_gp41; Stx2-converting phage 1717; verocytotoxin 2 variant 2c subunit A; K11006 shiga toxin subunit A
+vg:6972909 stx2cB, Stx2-1717_gp42; Stx2-converting phage 1717; verocytotoxin 2 variant 2c subunit B; K11007 shiga toxin subunit B
+vg:6973079 YYZ_gp39; Enterobacteria phage YYZ-2008; Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+vg:6973080 YYZ_gp40; Enterobacteria phage YYZ-2008; Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+vg:13828571 stx2A, D300_gp43; Escherichia phage P13374; shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:13828535 stx2B, D300_gp42; Escherichia phage P13374; shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:14005228 F366_gp36; Escherichia phage TL-2011c; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:14005229 F366_gp37; Escherichia phage TL-2011c; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:1262249 stx2A, VT2-Sap42; Enterobacteria phage VT2-Sakai; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:1262250 stx2B, VT2-Sap43; Enterobacteria phage VT2-Sakai; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/test-data/shiga-toxin-example-OR-broken
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/get_data/kegg_glycan/test-data/shiga-toxin-example-OR-broken Fri May 06 08:05:48 2016 -0400
@@ -0,0 +1,103 @@
+ece:Z1464 stx2A; shiga-like toxin II A subunit encoded by bacteriophage BP-933W; K11006 shiga toxin subunit A
+ece:Z1465 stx2B; shiga-like toxin II B subunit encoded by bacteriophage BP-933W; K11007 shiga toxin subunit B
+ece:Z3343 stx1B; shiga-like toxin 1 subunit B encoded within prophage CP-933V; K11007 shiga toxin subunit B
+ece:Z3344 stx1A; shiga-like toxin 1 subunit A encoded within prophage CP-933V; K11006 shiga toxin subunit A
+ecs:ECs1205 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+ecs:ECs1206 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+ecs:ECs2973 Shiga toxin I subunit B; K11007 shiga toxin subunit B
+ecs:ECs2974 Shiga toxin I subunit A; K11006 shiga toxin subunit A
+ecf:ECH74115_2905 shigatoxin 2, subunit B; K11007 shiga toxin subunit B
+ecf:ECH74115_2906 shiga toxin subunit A; K11006 shiga toxin subunit A
+ecf:ECH74115_3532 shiga toxin 2 B subunit; K11007 shiga toxin subunit B
+ecf:ECH74115_3533 shiga toxin subunit A; K11006 shiga toxin subunit A
+etw:ECSP_2722 stx2cB; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+etw:ECSP_2723 stx2A1; Shiga-like toxin II subunit A precursor; K11006 shiga toxin subunit A
+etw:ECSP_3252 stx2B; shiga toxin II subunit B; K11007 shiga toxin subunit B
+etw:ECSP_3253 stx2A2; shiga toxin II subunit A; K11006 shiga toxin subunit A
+elx:CDCO157_1154 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+elx:CDCO157_1155 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+elx:CDCO157_2738 Shiga toxin I subunit B precursor; K11007 shiga toxin subunit B
+elx:CDCO157_2739 Shiga toxin I subunit A precursor; K11006 shiga toxin subunit A
+eoj:ECO26_1599 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoj:ECO26_1600 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eoi:ECO111_2429 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eoi:ECO111_2430 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+eoi:ECO111_3361 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoi:ECO111_3362 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eoh:ECO103_2844 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eoh:ECO103_2845 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+eoh:ECO103_5197 Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+eoh:ECO103_5198 Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+eso:O3O_09170 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+eso:O3O_09165 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+esm:O3M_16105 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+esm:O3M_16110 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+esl:O3K_16130 Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+esl:O3K_16135 Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+ecoo:ECRM13514_3144 stx2B; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+ecoo:ECRM13514_3145 stx2A; Shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+ecoh:ECRM13516_2315 stx2B; Shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+ecoh:ECRM13516_2316 stx2A; Shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+stm:STM0284 shiga-like toxin A subunit
+seo:STM14_0332 putative Shiga-like toxin A subunit
+sev:STMMW_02891 putative shiga-like toxin A subunit
+sem:STMDT12_C02790 putative Shiga-like toxin A subunit
+sej:STMUK_0287 putative Shiga-like toxin A subunit
+setu:STU288_13325 putative shiga-like toxin A subunit
+setc:CFSAN001921_15985 Shiga toxin A subunit
+send:DT104_02831 sciR; shiga-like toxin a subunit, type VI secretion system SciR protein flags: precursor
+seni:CY43_01420 Shiga toxin A subunit
+seen:SE451236_07435 Shiga toxin A subunit
+spt:SPA2480 putative shiga-like toxin A subunit
+sek:SSPA2315 shiga-like toxin A subunit
+sei:SPC_0292 putative shiga-like toxin A subunit
+sec:SCH_0277 putative shiga-like toxin A subunit
+shb:SU5_0926 SciR Shiga-like toxin A subunit
+senh:CFSAN002069_07720 Shiga toxin A subunit
+seeh:SEEH1578_10515 putative shiga-like toxin A subunit
+seec:CFSAN002050_07960 Shiga toxin A subunit
+sene:IA1_01505 Shiga toxin A subunit
+senc:SEET0819_08350 Shiga toxin A subunit
+sdy:SDY_1389 stxA; Shiga toxin subunit A; K11006 shiga toxin subunit A
+sdy:SDY_1390 stxB; Shiga toxin subunit B; K11007 shiga toxin subunit B
+sdz:Asd1617_01826 Shiga toxin subunit A (EC:3.2.2.22); K11006 shiga toxin subunit A
+ebi:EbC_13780 Putative shiga-like toxin A subunit
+psos:POS17_2009 P9; putative Shiga-like toxin alpha subunit
+ssab:SSABA_v1c03870 hypothetical protein; K11006 shiga toxin subunit A
+ssab:SSABA_v1c09020 hypothetical protein; K11006 shiga toxin subunit A
+calt:Cal6303_2132 ribosome-inactivating protein; K11006 shiga toxin subunit A
+vg:22277199 PI30_gp34; Shigella phage POCJ13; shiga toxin A-chain precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:22277200 PI30_gp35; Shigella phage POCJ13; shiga toxin I subunit B precursor; K11007 shiga toxin subunit B
+vg:1262301 APSE-1_07; Endosymbiont phage APSE-1; P7; K11007 shiga toxin subunit B
+vg:1481747 stx1A, Stx1_p142; Escherichia Stx1 converting phage; Shiga toxin1 subunit A; K11006 shiga toxin subunit A
+vg:1481748 stx1B, Stx1_p143; Escherichia Stx1 converting phage; Shiga toxin1 subunit B; K11007 shiga toxin subunit B
+vg:1481767 stx2A, Stx2II_p143; Stx2 converting phage II; Shiga toxin2 subunit A; K11006 shiga toxin subunit A
+vg:1481768 stx2B, Stx2II_p144; Stx2 converting phage II; Shiga toxin2 subunit B; K11007 shiga toxin subunit B
+vg:26228052 AS349_gp84; Escherichia phage PA2; shiga-like toxin II subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:26228100 AS349_gp83; Escherichia phage PA2; shiga-like toxin II subunit B precursor; K11007 shiga toxin subunit B
+vg:26516496 stx2A, AU085_gp42; Enterobacteria phage VT2phi_272; Shiga toxin 2 A subunit; K11006 shiga toxin subunit A
+vg:26516448 stx2B, AU085_gp41; Enterobacteria phage VT2phi_272; Shiga toxin 2 B subunit; K11007 shiga toxin subunit B
+vg:26516283 stxB, phi191_00011, AU083_gp11; Escherichia phage phi191; shiga toxin subunit B precursor; K11007 shiga toxin subunit B
+vg:26516284 stxA, phi191_00012, AU083_gp12; Escherichia phage phi191; shiga toxin subunit A precursor (EC:3.2.2.22); K11006 shiga toxin subunit A
+vg:26519429 AU154_gp39; Shigella phage Ss-VASD; Stx1 A subunit; K11006 shiga toxin subunit A
+vg:26519430 AU154_gp40; Shigella phage Ss-VASD; Stx1 B subunit; K11007 shiga toxin subunit B
+vg:1261950 stxA2, L0103, sltIIA, stx2A, 933Wp40; Enterobacteria phage 933W; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:1262010 stxB2, L0104, sltIIB, stx2B, 933Wp41; Enterobacteria phage 933W; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:2641645 stxA1, phi4795p38, PBV4795_ORF40; Enterobacteria phage BP-4795; Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+vg:2641657 stxB1, phi4795p39, PBV4795_ORF41; Enterobacteria phage BP-4795; Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+vg:929695 stxA2e, P27p25; Enterobacteria phage phiP27; shiga toxin 2e A-subunit; K11006 shiga toxin subunit A
+vg:929727 stxB2e, P27p26; Enterobacteria phage phiP27; shiga toxin 2e B-subunit; K11007 shiga toxin subunit B
+vg:4397483 stx2A, Stx2-86_gp01; Stx2-converting phage 86; shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:4397484 stx2B, Stx2-86_gp02; Stx2-converting phage 86; shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:6159405 stx2A, pMIN27_41; Enterobacteria phage Min27; ORF44; K11006 shiga toxin subunit A
+vg:6159351 stx2B, pMIN27_42; Enterobacteria phage Min27; ORF45; K11007 shiga toxin subunit B
+vg:6973138 Stx2-1717_gp41; Stx2-converting phage 1717; verocytotoxin 2 variant 2c subunit A; K11006 shiga toxin subunit A
+vg:6972909 stx2cB, Stx2-1717_gp42; Stx2-converting phage 1717; verocytotoxin 2 variant 2c subunit B; K11007 shiga toxin subunit B
+vg:6973079 YYZ_gp39; Enterobacteria phage YYZ-2008; Shiga toxin 1 subunit A; K11006 shiga toxin subunit A
+vg:6973080 YYZ_gp40; Enterobacteria phage YYZ-2008; Shiga toxin 1 subunit B; K11007 shiga toxin subunit B
+vg:13828571 stx2A, D300_gp43; Escherichia phage P13374; shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:13828535 stx2B, D300_gp42; Escherichia phage P13374; shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:14005228 F366_gp36; Escherichia phage TL-2011c; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:14005229 F366_gp37; Escherichia phage TL-2011c; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
+vg:1262249 stx2A, VT2-Sap42; Enterobacteria phage VT2-Sakai; Shiga toxin 2 subunit A; K11006 shiga toxin subunit A
+vg:1262250 stx2B, VT2-Sap43; Enterobacteria phage VT2-Sakai; Shiga toxin 2 subunit B; K11007 shiga toxin subunit B
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/test_findKEGG.py
--- a/get_data/kegg_glycan/test_findKEGG.py Wed Mar 23 14:35:56 2016 -0400
+++ b/get_data/kegg_glycan/test_findKEGG.py Fri May 06 08:05:48 2016 -0400
@@ -41,9 +41,10 @@
m = fk.find_entries_in_KEGG("john", "glucose")
def test_unfindable_entry(self):
- """should return None"""
+ """should return an empty string"""
m = fk.find_entries_in_KEGG("glycan", "sally")
- self.assertIsNone(m)
+ emptystring=""
+ self.assertItemsEqual(m,emptystring,"Expected Empty String for non-existent entry")
def test_find_example(self):
m = fk.find_entries_in_KEGG("glycan", "glucose")
@@ -52,9 +53,123 @@
def test_query_has_newlines(self):
"""
- should be fine and return entries
+ Assume a new line query is an AND query.
+ """
+ m = fk.find_entries_in_KEGG("glycan", "glucose\nUDP")
+ self.assertIsNotNone(m)
+ print m
+ self.assertIn("UDP-D-glucose", m)
+
+ def test_query_has_newlines_and_a_space(self):
+ """
+ glucose\\n UDP. OR's should pass
"""
m = fk.find_entries_in_KEGG("glycan", "glucose\n UDP")
self.assertIsNotNone(m)
self.assertIn("UDP-D-glucose", m)
+ def test_gene_specific_or(self):
+ """
+ test "shiga toxin" returns.
+ /find/genes/"shiga toxin" for keywords "shiga toxin"
+ """
+ m = fk.find_entries_in_KEGG("genes","shiga toxin")
+ self.assertIsNotNone(m)
+ #. not matching to all entries - that is silly, the db get updated!
+ self.assertIn("shiga", m)
+ self.assertIn("toxin", m)
+ self.assertNotIn("stm:STM0284",m) # should not be found in the OR query
+
+ def test_gene_specific_and(self):
+ """
+ test "shiga+toxin" returns. it should
+ /find/genes/shiga+toxin for keywords "shiga" and "toxin"
+ """
+ m = fk.find_entries_in_KEGG("genes","shiga+toxin")
+ self.assertIsNotNone(m)
+ #. not matching to all entries - that is silly, the db get updated!
+ self.assertIn("shiga", m)
+ self.assertIn("toxin", m)
+ self.assertIn("stm:STM0284",m) # should be found in the AND query
+
+ def test_enzyme_specific_1(self):
+ """
+ test 2.4.99.1 returns
+ """
+ m = fk.find_entries_in_KEGG("enzyme","2.4.99.1")
+ self.assertIsNotNone(m)
+ #. not matching to all entries - that is silly, the db get updated!
+ self.assertIn("2.4.99.1", m)
+
+ def test_enzyme_specific_2(self):
+ """
+ test 2.4.99.6 returns
+ """
+ m = fk.find_entries_in_KEGG("enzyme","2.4.99.6")
+ self.assertIsNotNone(m)
+ #. not matching to all entries - that is silly, the db get updated!
+ self.assertIn("2.4.99.6", m)
+
+ def test_enzyme_specific_3(self):
+ """
+ test ec: 2.4.99.6 with space (tests stripping space functionality)
+ """
+ m = fk.find_entries_in_KEGG("enzyme","ec: 2.4.99.6")
+ self.assertIsNotNone(m)
+ #. not matching to all entries - that is silly, the db get updated!
+ self.assertIn("2.4.99.6", m)
+
+ def test_enzyme_specific_4(self):
+ """
+ test ec: 2.4.99. , should return 2.4.99.*
+ """
+ m = fk.find_entries_in_KEGG("enzyme","ec: 2.4.99.")
+ self.assertIsNotNone(m)
+ #. not matching to all entries - that is silly, the db get updated!
+ self.assertIn("2.4.99.1", m)
+ self.assertIn("2.4.99.6", m)
+ self.assertIn("2.4.99.11", m)
+
+ def test_enzyme_specific_5(self):
+ """
+ test ec: 2.4.99.1+2.4.99.6 . AND function, so returns nothing here. Do not expect an enzyme to be doubly classified.
+ """
+ m = fk.find_entries_in_KEGG("enzyme","2.4.99.1+2.4.99.6")
+ self.assertIsNotNone(m)
+ emptystring=""
+ self.assertItemsEqual(m,emptystring,"Expected Empty String for non-existent entry")
+
+ def test_enzyme_specific_6(self):
+ """
+ test ec: 2.4.99.1 2.4.99.6 . OR function, should return something but does not (KEGG FAILURE?).
+ note that trying this without "" will return the AND function result which is incorrect.
+ """
+ m = fk.find_entries_in_KEGG("enzyme","2.4.99.1 2.4.99.6")
+ self.assertIsNotNone(m)
+ #self.assertIn("2.4.99.1", m)
+ emptystring=""
+ print m
+ self.assertItemsEqual(m,emptystring,"Expected Empty String for non-existent entry")
+
+
+ def test_enzyme_specific_7(self):
+ """
+ test deoxy+1.1.1. . AND function, that should return something
+ """
+ m = fk.find_entries_in_KEGG("enzyme","deoxy+1.1.1")
+ self.assertIsNotNone(m)
+ print m
+ self.assertIn("deoxy", m)
+ self.assertIn("1.1.1", m)
+
+ def test_enzyme_specific_8(self):
+ """
+ test deoxy 1.1.1. . OR function. should return deoxy or 1.1.1. but does not (KEGG FAILURE?). i
+ note that trying this without "" will return the AND function result which is incorrect.
+ """
+ m = fk.find_entries_in_KEGG("enzyme","deoxy 1.1.1")
+ self.assertIsNotNone(m)
+ emptystring=""
+ print m
+ self.assertItemsEqual(m,emptystring,"Expected Empty String for non-existent entry")
+
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/test_findKEGG.pyc
Binary file get_data/kegg_glycan/test_findKEGG.pyc has changed
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/test_getkcfKEGG.pyc
Binary file get_data/kegg_glycan/test_getkcfKEGG.pyc has changed
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/test_linkKEGG.py
--- a/get_data/kegg_glycan/test_linkKEGG.py Wed Mar 23 14:35:56 2016 -0400
+++ b/get_data/kegg_glycan/test_linkKEGG.py Fri May 06 08:05:48 2016 -0400
@@ -40,3 +40,76 @@
m = lk.linked_entries_from_kegg("br", "gl:G10496")
self.assertEquals(m,"")
+ def test_enzyme_glycan_search_1(self):
+ """
+ test "2.4.99.1 ec: " returns
+ """
+ m = lk.linked_entries_from_kegg("glycan", "2.4.99.1 ec:")
+ self.assertIn("2.4.99.1", m)
+
+
+ def test_enzyme_glycan_search_2(self):
+ """
+ test "2.4.99.1 ec: 2.4.99.6" returns
+ """
+ m = lk.linked_entries_from_kegg("glycan", "2.4.99.1 ec: 2.4.99.6" )
+ self.assertIn("2.4.99.1", m)
+ self.assertIn("2.4.99.6", m)
+
+ def test_enzyme_glycan_search_3(self):
+ """
+ test "2.4.99.1 ec:2.4.99.6" returns
+ """
+ m = lk.linked_entries_from_kegg("glycan", "2.4.99.1 ec:2.4.99.6" )
+ self.assertIn("2.4.99.1", m)
+ self.assertIn("2.4.99.6", m)
+
+ def test_enzyme_glycan_search_4(self):
+ """
+ test "2.4.99.1+2.4.99.6" returns. This time '+' is or.
+ """
+ m = lk.linked_entries_from_kegg("glycan", "2.4.99.1+2.4.99.6" )
+ self.assertIn("2.4.99.1", m)
+ self.assertIn("2.4.99.6", m)
+
+ def test_enzyme_glycan_search_4b(self):
+ """
+ test "2.4.99.1+ 2.4.99.6" returns. This time '+' is or.
+ """
+ m = lk.linked_entries_from_kegg("glycan", "2.4.99.1+ 2.4.99.6" )
+ self.assertIn("2.4.99.1", m)
+ self.assertIn("2.4.99.6", m)
+
+ def test_enzyme_glycan_search_5(self):
+ """
+ test "2.4.99.1 2.4.99.6" returns. Space also means or for the link db.
+ """
+ m = lk.linked_entries_from_kegg("glycan", "2.4.99.1 2.4.99.6" )
+ self.assertIn("2.4.99.1", m)
+ self.assertIn("2.4.99.6", m)
+
+
+ def test_enzyme_glycan_search_6(self):
+ """
+ test "2.4.99.1+nana+2.4.99.6" returns. strangely when inserting junk it still works.
+ """
+ m = lk.linked_entries_from_kegg("glycan", "2.4.99.1+nana+2.4.99.6" )
+ self.assertIn("2.4.99.1", m)
+ self.assertIn("2.4.99.6", m)
+
+ def test_enzyme_glycan_search_6b(self):
+ """
+ test "2.4.99.1 nana 2.4.99.6" returns. strangely when inserting junk it still works.
+ """
+ m = lk.linked_entries_from_kegg("glycan", "2.4.99.1+nana+2.4.99.6" )
+ self.assertIn("2.4.99.1", m)
+ self.assertIn("2.4.99.6", m)
+
+ def test_enzyme_glycan_search_7(self):
+ """
+ test "\"2.4.99.1\"" is a bad request as it includes quotes.
+ """
+ from urllib2 import HTTPError
+ with self.assertRaises(HTTPError):
+ m = lk.linked_entries_from_kegg("glycan", "\"2.4.99.1\"" )
+
diff -r 89592faa2875 -r 0a5e0df17054 get_data/kegg_glycan/test_linkKEGG.pyc
Binary file get_data/kegg_glycan/test_linkKEGG.pyc has changed