annotate EnzymeEntry.R @ 2:20d69a062da3 draft

planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
author prog
date Thu, 02 Mar 2017 08:55:00 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
1 #####################
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
2 # CLASS DECLARATION #
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
3 #####################
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
4
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
5 EnzymeEntry <- methods::setRefClass("EnzymeEntry", contains = 'BiodbEntry')
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
6
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
7 ###########
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
8 # FACTORY #
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
9 ###########
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
10
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
11 createEnzymeEntryFromTxt <- function(contents, drop = TRUE) {
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
12
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
13 entries <- list()
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
14
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
15 # Define fields regex
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
16 regex <- character()
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
17 regex[[BIODB.ACCESSION]] <- "^ID\\s+([0-9.]+)$"
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
18 regex[[BIODB.DESCRIPTION]] <- "^DE\\s+(.+)$"
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
19
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
20 for (text in contents) {
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
21
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
22 # Create instance
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
23 entry <- EnzymeEntry$new()
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
24
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
25 lines <- strsplit(text, "\n")
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
26 for (s in lines[[1]]) {
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
27
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
28 # Test generic regex
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
29 parsed <- FALSE
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
30 for (field in names(regex)) {
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
31 g <- stringr::str_match(s, regex[[field]])
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
32 if ( ! is.na(g[1,1])) {
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
33 entry$setField(field, g[1,2])
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
34 parsed <- TRUE
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
35 break
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
36 }
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
37 }
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
38 if (parsed)
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
39 next
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
40 }
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
41
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
42 entries <- c(entries, entry)
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
43 }
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
44
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
45 # Replace elements with no accession id by NULL
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
46 entries <- lapply(entries, function(x) if (is.na(x$getField(BIODB.ACCESSION))) NULL else x)
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
47
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
48 # If the input was a single element, then output a single object
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
49 if (drop && length(contents) == 1)
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
50 entries <- entries[[1]]
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
51
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
52 return(entries)
20d69a062da3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8
prog
parents:
diff changeset
53 }