Fixing contaminant database"
In prozor: Minimal Protein Set Explaining Peptide Spectrum Matches

Look at Library

rm(list = ls())
library(prozor)
library(quantable)

prots <-
    readPeptideFasta("inst/extdata/fgcz_contaminants_20130115_patchedWithRespectiveUniprot2.fasta")
length(prots)

tmp <- lapply(prots, function(x) {
    attributes(x)$Annot
})
tmp <- cbind(names = names(tmp), ids = unlist(tmp))

xx <- split2table(tmp[, "ids"], split = "\\| eValue")

ids <- grep("sp\\|", xx[, 2])
length(ids)
xxWithAnnot <- xx[ids, ]
blastpositive <- split2table(xxWithAnnot[, 2], split = "sp\\|")
blastpositive

blasthom <- gsub(": ", "", blastpositive[, 1])
library(stringr)
blasthom <- str_trim(blasthom)
blasthom <- data.frame(split2table(blasthom, split = " "))
head(blasthom)
blasthom <- do.call(paste, c(blasthom[, 2:1], sep = "_"))
conts <- xx[ids, 1]
conts <- gsub("ZZ_", "", gsub(">", "", conts))

finids <-
    do.call(paste, c(data.frame(conts, blastpositive[, 2]), sep = "_"))

Annot <- do.call(paste, c(data.frame(finids, blasthom), sep = " "))
length(Annot)


noAnnot <- xx[-ids, ]

noAnnot <- gsub("ZZ_", "", (noAnnot))

noAnnot <- gsub("^>", "", (noAnnot))
noAnnot[31, 1] <-
    gsub("\\| iRT protein with AAAA spacer", "", noAnnot[31, 1])
noAnnot[31, 2] <- gsub("zz\\|FGCZCont0260\\| ", "", noAnnot[31, 2])

noAnnot <-
    gsub(": 0.0 blastpHomologue ", "ambigous_blastpHomologue", noAnnot)

noAnnot
tt <- gsub("zz\\|", "", noAnnot[, 1])
xt <- do.call(paste, c(data.frame(noAnnot[, 1], tt), sep = "_"))
noAnnot[, 1] <- do.call(paste, c(data.frame(xt , tt), sep = "|"))
noAnnot <- do.call(paste, c(data.frame(noAnnot), sep = " "))
length(ids)
length(Annot)
res <- rep(NA, nrow(xx))
res[ids] <- Annot
res[-ids] <- noAnnot

ids <- split2table(res, split = " [a-z]")[, 1]
library(seqinr)

aseq <-
    function(id, seq, annot) {
        return(as.SeqFastaAA(seq, name = id, Annot = paste(">", annot, sep = "")))
    }
aseq(ids[1], as.character(prots[1]), res[1])
tmp <- mapply(aseq, ids , as.character(prots) , res  , SIMPLIFY = FALSE)

writeFasta(tmp, file = "inst/extdata/fgcz_ContaminantsWithAnnotation.fasta")


dummy <-
    as.SeqFastaAA("CRAPCRAPCRAP", name = "sp|Q15323|K1H1_HUMAN", Annot = ">zz|ZZ_FGCZCont0005| eValue: 1.0e-54 blastpHomologue sp|Q15323|K1H1_HUMAN")
dummy

">zz|FGCZCont0000_P61626|LYSC_HUMAN bestBlastpHomologue_0.0"

prots <-
    readPeptideFasta("inst/extdata/fgcz_ContaminantsWithAnnotation.fasta")
length(prots)

nonhuman <- prots[!grepl("_HUMAN", names(prots))]
writeFasta(nonhuman, file = "inst/extdata/fgcz_ContaminantsWithAnnotationNoHuman.fasta")