Mercurial > repos > prog > lcmsmatching

annotate UniProtCompound.R @ 0:e66bb061af06 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
author prog
date Tue, 12 Jul 2016 12:02:37 -0400
parents
children 253d531a0193
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
1 if ( ! exists('UniprotCompound')) { # Do not load again if already loaded
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
2
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
3 source('BiodbEntry.R')
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
4
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
5 #####################
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
6 # CLASS DECLARATION #
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
7 #####################
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
8
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
9 UniprotCompound <- setRefClass("UniprotCompound", contains = "BiodbEntry")
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
10
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
11 ###########
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
12 # FACTORY #
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
13 ###########
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
14
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
15 createUniprotCompoundFromXml <- function(contents, drop = FALSE) {
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
16
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
17 library(XML)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
18
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
19 # Set XML namespace
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
20 ns <- c(uniprot = "http://uniprot.org/uniprot")
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
21
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
22 compounds <- list()
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
23
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
24 # Define xpath expressions
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
25 xpath.values <- character()
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
26 xpath.values[[RBIODB.NAME]] <- "/uniprot:uniprot/uniprot:compound/uniprot:name"
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
27 xpath.values[[RBIODB.GENE.SYMBOLS]] <- "//uniprot:gene/uniprot:name"
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
28 xpath.values[[RBIODB.FULLNAMES]] <- "//uniprot:protein//uniprot:fullName"
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
29 xpath.values[[RBIODB.SEQUENCE]] <- "//uniprot:entry/uniprot:sequence"
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
30 xpath.values[[RBIODB.ACCESSION]] <- "//uniprot:accession[1]"
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
31 xpath.attr <- list()
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
32 xpath.attr[[RBIODB.KEGG.ID]] <- list(path = "//uniprot:dbReference[@type='KEGG']", attr = 'id')
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
33 xpath.attr[[RBIODB.NCBI.GENE.ID]] <- list(path = "//uniprot:dbReference[@type='GeneID']", attr = 'id')
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
34 xpath.attr[[RBIODB.ENZYME.ID]] <- list(path = "//uniprot:dbReference[@type='EC']", attr = 'id')
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
35 xpath.attr[[RBIODB.MASS]] <- list(path = "//uniprot:entry/uniprot:sequence", attr = 'mass')
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
36 xpath.attr[[RBIODB.LENGTH]] <- list(path = "//uniprot:entry/uniprot:sequence", attr = 'length')
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
37
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
38 for (content in contents) {
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
39
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
40 # Create instance
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
41 compound <- HmdbCompound$new()
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
42
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
43 # If the entity doesn't exist (i.e.: no <id>.xml page), then it returns an HTML page
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
44 if ( ! grepl("^<!DOCTYPE html ", content, perl = TRUE)) {
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
45
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
46 # Parse XML
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
47 xml <- xmlInternalTreeParse(content, asText = TRUE)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
48
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
49 # Test value xpath
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
50 for (field in names(xpath.values)) {
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
51 v <- xpathSApply(xml, xpath.values[[field]], xmlValue, namespaces = ns)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
52 if (length(v) > 0)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
53 compound$setField(field, v)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
54 }
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
55
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
56 # Test attribute xpath
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
57 for (field in names(xpath.attr)) {
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
58 v <- xpathSApply(xml, xpath.attr[[field]]$path, xmlGetAttr, xpath.attr[[field]]$attr, namespaces = ns)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
59 if (length(v) > 0)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
60 compound$setField(field, v)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
61 }
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
62
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
63 # Remove new lines from sequence string
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
64 seq <- compound$getField(RBIODB.SEQUENCE)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
65 if ( ! is.na(seq))
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
66 compound$setField(RBIODB.SEQUENCE, gsub("\\n", "", seq))
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
67 }
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
68
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
69 compounds <- c(compounds, compound)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
70 }
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
71
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
72 # Replace elements with no accession id by NULL
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
73 compounds <- lapply(compounds, function(x) if (is.na(x$getField(RBIODB.ACCESSION))) NULL else x)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
74
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
75 # If the input was a single element, then output a single object
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
76 if (drop && length(contents) == 1)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
77 compounds <- compounds[[1]]
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
78
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
79 return(compounds)
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
80 }
e66bb061af06 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 3529b25417f8e1a5836474c9adec4b696d35099d-dirty
prog
parents:
diff changeset
81 }