Mercurial > repos > prog > lcmsmatching

annotate ChemspiderConn.R @ 1:45e985cd8e9e 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 d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
author prog
date Tue, 31 Jan 2017 05:27:24 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
1 #####################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
2 # CLASS DECLARATION #
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
3 #####################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
4
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
5 ChemspiderConn <- methods::setRefClass("ChemspiderConn", contains = "RemotedbConn")
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
6
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
7 ##########################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
8 # GET ENTRY CONTENT TYPE #
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
9 ##########################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
10
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
11 ChemspiderConn$methods( getEntryContentType = function() {
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
12 return(BIODB.XML)
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
13 })
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
14
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
15 #####################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
16 # GET ENTRY CONTENT #
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
17 #####################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
18
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
19 ChemspiderConn$methods( getEntryContent = function(ids) {
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
20
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
21 # Debug
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
22 .self$.print.debug.msg(paste0("Get entry content(s) for ", length(ids)," id(s)..."))
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
23
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
24 URL.MAX.LENGTH <- 2083
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
25
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
26 # Initialize return values
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
27 content <- rep(NA_character_, length(ids))
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
28
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
29 # Loop on all
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
30 n <- 0
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
31 inc <- NA_integer_
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
32 while (n < length(ids)) {
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
33
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
34 # Get list of accession ids to retrieve
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
35 accessions <- ids[(n + 1):(if (is.na(inc)) length(ids) else (min(n + inc, length(ids))))]
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
36
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
37 # Create URL request
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
38 x <- get.entry.url(class = BIODB.CHEMSPIDER, accession = accessions, content.type = BIODB.XML, max.length = URL.MAX.LENGTH, base.url = .self$.url, token = .self$.token)
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
39
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
40 # Debug
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
41 .self$.print.debug.msg(paste0("Send URL request for ", x$n," id(s)..."))
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
42
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
43 # Send request
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
44 xmlstr <- .self$.get.url(x$url)
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
45
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
46 # Error : "Cannot convert WRONG to System.Int32.\r\nParameter name: type ---> Input string was not in a correct format.\r\n"
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
47 if (grepl('^Cannot convert .* to System\\.Int32\\.', xmlstr)) {
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
48 # One of the ids is incorrect
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
49 if (is.na(inc)) {
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
50 inc <- 1
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
51 next
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
52 }
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
53 else
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
54 xmlstr <- NA_character_
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
55 }
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
56
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
57 # Increase number of entries retrieved
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
58 n <- n + x$n
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
59
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
60 # Parse XML and get included XML
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
61 if ( ! is.na(xmlstr)) {
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
62 xml <- xmlInternalTreeParse(xmlstr, asText = TRUE)
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
63 ns <- c(csns = "http://www.chemspider.com/")
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
64 returned.ids <- xpathSApply(xml, "//csns:ExtendedCompoundInfo/csns:CSID", xmlValue, namespaces = ns)
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
65 content[match(returned.ids, ids)] <- vapply(getNodeSet(xml, "//csns:ExtendedCompoundInfo", namespaces = ns), saveXML, FUN.VALUE = '')
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
66 }
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
67
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
68 # Debug
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
69 .self$.print.debug.msg(paste0("Now ", length(ids) - n," id(s) left to be retrieved..."))
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
70 }
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
71
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
72 return(content)
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
73 })
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
74
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
75 ################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
76 # CREATE ENTRY #
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
77 ################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
78
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
79 ChemspiderConn$methods( createEntry = function(content, drop = TRUE) {
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
80 return(createChemspiderEntryFromXml(content, drop = drop))
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
81 })
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
82
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
83 ############################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
84 # GET CHEMSPIDER IMAGE URL #
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
85 ############################
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
86
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
87 get.chemspider.image.url <- function(id) {
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
88
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
89 url <- paste0('http://www.chemspider.com/ImagesHandler.ashx?w=300&h=300&id=', id)
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
90
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
91 return(url)
45e985cd8e9e planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8-dirty
prog
parents:
diff changeset
92 }