Merge pull request #36 from mdnestor/dev

Added is_contaminant column to MoTrPAC tables

DESCRIPTION

@@ -21,7 +21,11 @@ Imports:
    data.table,
    dplyr,
    plyr,
+   IRanges,
+   MSnbase,
    odbc,
+   purrr,
+   tibble,
    tidyr,
    tidyselect,
    tools,

R/motrpac_bic_funtions.R

@@ -83,7 +83,7 @@ make_rii_peptide_gl <- function(msnid, masic_data, fractions, samples,
   ## Create RII peptide table
   rii_peptide <- crosstab %>%
     select(Specie) %>%
-    mutate(protein_id = sub("(^.*\\.\\d+)@.*", "\\1", Specie),
+    mutate(protein_id = sub("(^.*)@(.*)", "\\1", Specie),
            sequence = sub("(^.*)@(.*)", "\\2", Specie),
            organism_name = org_name) %>%
     mutate(REFSEQ = sub("(^.*)\\.\\d+", "\\1", protein_id))
@@ -95,19 +95,20 @@ make_rii_peptide_gl <- function(msnid, masic_data, fractions, samples,
     inner_join(., conv)
 
   rii_peptide <- rii_peptide %>%
-    inner_join(conv) %>%
+    left_join(conv) %>%
     rename(gene_symbol = SYMBOL,
            entrez_id = ENTREZID) %>%
     select(-REFSEQ)
 
   ## Additional info from MS/MS
   ids <- psms(msnid) %>%
-    select(accession, peptide, redundantAccessions, MSGFDB_SpecEValue) %>%
+    select(accession, peptide, noninferableProteins, MSGFDB_SpecEValue) %>%
     rename(protein_id = accession,
            sequence = peptide,
-           redundant_ids = redundantAccessions) %>%
+           redundant_ids = noninferableProteins) %>%
     group_by(protein_id, sequence, redundant_ids) %>%
-    summarize(peptide_score = min(MSGFDB_SpecEValue))
+    summarize(peptide_score = min(MSGFDB_SpecEValue)) %>%
+    mutate(is_contaminant = grepl("Contaminant", protein_id))
 
   rii_peptide <- inner_join(rii_peptide, ids)
 
@@ -128,6 +129,10 @@ make_results_ratio_gl <- function(msnid, masic_data, fractions, samples,
   crosstab <- create_crosstab(msnid, masic_data, aggregation_level, fractions,
                               samples, references)
 
+  # testing purposes
+  #crosstab <- rbind(crosstab, rnorm(10))
+  #rownames(crosstab)[nrow(crosstab)] <- "Contaminant_TRYP_PIG"
+
   crosstab <- as.data.frame(crosstab) %>%
     rownames_to_column("protein_id")
 
@@ -144,7 +149,7 @@ make_results_ratio_gl <- function(msnid, masic_data, fractions, samples,
     inner_join(., conv)
 
   results_ratio <- results_ratio %>%
-    inner_join(conv) %>%
+    left_join(conv) %>%
     rename(gene_symbol = SYMBOL,
            entrez_id = ENTREZID) %>%
     select(-REFSEQ)
@@ -162,7 +167,8 @@ make_results_ratio_gl <- function(msnid, masic_data, fractions, samples,
     summarize(peptide_score = min(MSGFDB_SpecEValue)) %>%
     group_by(protein_id, redundant_ids, percent_coverage) %>%
     summarize(protein_score = min(peptide_score),
-              num_peptides = n()) 
+              num_peptides = n()) %>%
+    mutate(is_contaminant = grepl("Contaminant", protein_id))
 
   results_ratio <- inner_join(results_ratio, ids)
 
@@ -214,7 +220,7 @@ make_rii_peptide_ph <- function(msnid, masic_data, fractions, samples, reference
   conv <- fetch_conversion_table(org_name, from = "REFSEQ", "ENTREZID") %>%
     inner_join(., conv)
 
-  rii_peptide <- inner_join(rii_peptide, conv) %>%
+  rii_peptide <- left_join(rii_peptide, conv) %>%
     rename(gene_symbol = SYMBOL,
            entrez_id = ENTREZID) %>%
     select(-REFSEQ)
@@ -233,7 +239,8 @@ make_rii_peptide_ph <- function(msnid, masic_data, fractions, samples, reference
     summarize(peptide_score = min(MSGFDB_SpecEValue),
               confident_score = max(maxAScore)) %>%
     mutate(confident_site = case_when(confident_score >= 17 ~ TRUE,
-                                      confident_score < 17 ~ FALSE))
+                                      confident_score < 17 ~ FALSE),
+           is_contaminant = grepl("Contaminant", protein_id))
 
   rii_peptide <- inner_join(rii_peptide, ids) %>%
     mutate(ptm_id = gsub("-", sep, ptm_id))
@@ -270,7 +277,7 @@ make_results_ratio_ph <- function(msnid, masic_data, fractions, samples,
   conv <- fetch_conversion_table(org_name, from = "REFSEQ", "ENTREZID") %>%
     inner_join(., conv)
 
-  results_ratio <- inner_join(results_ratio, conv) %>%
+  results_ratio <- left_join(results_ratio, conv) %>%
     rename(gene_symbol = SYMBOL,
            entrez_id = ENTREZID) %>%
     select(-REFSEQ)
@@ -294,7 +301,8 @@ make_results_ratio_ph <- function(msnid, masic_data, fractions, samples,
     summarize(ptm_score = min(peptide_score),
               confident_score = max(confident_score)) %>%
     mutate(confident_site = case_when(confident_score >= 17 ~ TRUE,
-                                      confident_score < 17 ~ FALSE))
+                                      confident_score < 17 ~ FALSE),
+           is_contaminant = grepl("Contaminant", protein_id))
 
   results_ratio <- inner_join(results_ratio, ids) %>%
     mutate(ptm_id = gsub("-", sep, ptm_id))