Mercurial > repos > prog > lcmsmatching
annotate ChebiCompound.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 |
| rev | line source |
|---|---|
|
0
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
1 if ( ! exists('ChebiCompound')) { # 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 ChebiCompound <- setRefClass("ChebiCompound", 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 createChebiCompoundFromHtml <- 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 # xpath.expr[[RBIODB.ACCESSION]] <- "//b[starts-with(., 'CHEBI:')]" |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
24 xpath.expr[[RBIODB.INCHI]] <- "//td[starts-with(., 'InChI=')]" |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
25 xpath.expr[[RBIODB.INCHIKEY]] <- "//td[text()='InChIKey']/../td[2]" |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
26 |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
27 for (html in contents) { |
|
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 # Create instance |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
30 compound <- ChebiCompound$new() |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
31 |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
32 # Parse HTML |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
33 xml <- htmlTreeParse(html, asText = TRUE, useInternalNodes = TRUE) |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
34 |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
35 # Test generic xpath expressions |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
36 for (field in names(xpath.expr)) { |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
37 v <- xpathSApply(xml, xpath.expr[[field]], xmlValue) |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
38 if (length(v) > 0) |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
39 compound$setField(field, v) |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
40 } |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
41 |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
42 # Get accession |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
43 accession <- xpathSApply(xml, "//b[starts-with(., 'CHEBI:')]", xmlValue) |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
44 if (length(accession) > 0) { |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
45 accession <- sub('^CHEBI:([0-9]+)$', '\\1', accession, perl = TRUE) |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
46 compound$setField(RBIODB.ACCESSION, accession) |
|
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 compounds <- c(compounds, compound) |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
50 } |
|
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 # 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
|
53 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
|
54 |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
55 # 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
|
56 if (drop && length(contents) == 1) |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
57 compounds <- compounds[[1]] |
|
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 return(compounds) |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
60 } |
|
3afe41d3e9e7
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit bb4d3e23d99828bfee16d31d794c49a17313ec2f
prog
parents:
diff
changeset
|
61 } |
