annotate ChemspiderCompound.R @ 0:3afe41d3e9e7 draft

planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
author prog
date Mon, 11 Jul 2016 09:12:03 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
1 if ( ! exists('ChemspiderCompound')) { # Do not load again if already loaded
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
2
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
3 source('BiodbEntry.R')
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
4
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
5 #####################
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
6 # CLASS DECLARATION #
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
7 #####################
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
8
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
9 ChemspiderCompound <- setRefClass("ChemspiderCompound", contains = "BiodbEntry")
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
10
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
11 ###########
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
12 # FACTORY #
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
13 ###########
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
14
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
15 createChemspiderCompoundFromHtml <- function(contents, drop = TRUE) {
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
16
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
17 library(XML)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
18
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
19 compounds <- list()
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
20
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
21 # Define xpath expressions
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
22 xpath.expr <- character()
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
23
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
24 for (html in contents) {
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
25
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
26 # Create instance
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
27 compound <- ChemspiderCompound$new()
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
28
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
29 # Parse HTML
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
30 xml <- htmlTreeParse(html, asText = TRUE, useInternalNodes = TRUE)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
31
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
32 # Test generic xpath expressions
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
33 for (field in names(xpath.expr)) {
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
34 v <- xpathSApply(xml, xpath.expr[[field]], xmlValue)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
35 if (length(v) > 0)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
36 compound$setField(field, v)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
37 }
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
38
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
39 # Get accession
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
40 accession <- xpathSApply(xml, "//li[starts-with(., 'ChemSpider ID')]", xmlValue)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
41 if (length(accession) > 0) {
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
42 accession <- sub('^ChemSpider ID([0-9]+)$', '\\1', accession, perl = TRUE)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
43 compound$setField(RBIODB.ACCESSION, accession)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
44 }
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
45
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
46 compounds <- c(compounds, compound)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
47 }
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
48
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
49 # Replace elements with no accession id by NULL
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
50 compounds <- lapply(compounds, function(x) if (is.na(x$getField(RBIODB.ACCESSION))) NULL else x)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
51
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
52 # If the input was a single element, then output a single object
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
53 if (drop && length(contents) == 1)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
54 compounds <- compounds[[1]]
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
55
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
56 return(compounds)
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
57 }
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
58 }
3afe41d3e9e7 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff changeset
59