release 1.0.9

DESCRIPTION

@@ -3,7 +3,7 @@ Package: msdap
 Title: Mass Spectrometry Downstream Analysis Pipeline
 Description: Analyze label-free proteomics datasets from various sources (MaxQuant, Spectronaut, etc) using a pipeline that facilitates peptide filtering and many algorithms for normalization and statistical analysis. A comprehensive PDF report can be generated that includes many data visualizations and documentation thereof.
 URL: https://github.com/ftwkoopmans/msdap
-Version: 1.0.8
+Version: 1.0.9
 Authors@R: 
     person(given = "Frank",
            family = "Koopmans",

NAMESPACE

@@ -2,6 +2,7 @@
 
 export(analysis_quickstart)
 export(append_log)
+export(as_matrix_except_first_column)
 export(cache_filtering_data)
 export(check_dataset_hascache)
 export(check_dataset_integrity)
@@ -49,6 +50,7 @@ export(initialize_multiprocessing)
 export(invalidate_cache)
 export(is_dia_dataset)
 export(logger.default)
+export(matrix_to_long)
 export(merge_fractionated_samples)
 export(merge_replicate_samples)
 export(mgi_lookuptable)

R/dea.R

@@ -17,13 +17,11 @@
 #'
 #' **msqrob**: implementation of the MSqRob package, with minor tweak for (situationally) faster computation (PMID:26566788) <https://github.com/statOmics/msqrob>. This is a peptide-level DEA algorithm. This method will take provided covariates into account (if any). Implemented in function; \code{de_msqrobsum_msqrob}
 #'
-#' **msqrob_fc**: msqrob estimated protein pvalues, but with log2fc values computed by eBayes. This avoids the excessive shrinkage of protein log2fc values by MSqRob, but does use the MSqRob robust peptide regression model for protein p-value estimates (which outperforms eBayes/DEqMS in ROC analyses on most benchmark datasets, so this method is intended as a "best of both worlds")
-#'
 #' **msqrobsum**: implementation of the MSqRob package (which also features MSqRobSum), with minor tweak for (situationally) faster computation (PMID:32321741) <https://github.com/statOmics/msqrob>. This is a hybrid peptide&protein-level DEA algorithm that takes peptide-level data as input; it first performs peptide-to-protein rollup, then applies statistics to this protein-level data matrix. This method will take provided covariates into account (if any). Implemented in function; \code{de_msqrobsum_msqrob}
 #'
 #' @export
 dea_algorithms = function() {
-  return(c("ebayes", "deqms", "msempire", "msqrobsum", "msqrob")) # , "msqrob_fc"
+  return(c("ebayes", "deqms", "msempire", "msqrobsum", "msqrob"))
 }
 
 

R/filter_peptides.R

@@ -651,8 +651,8 @@ normalize_peptide_intensity_column = function(dataset, col_intensity, norm_algor
   if(length(col_intensity) != 1 || !is.character(col_intensity) || ! col_intensity %in% colnames(dataset$peptides)) {
     append_log(sprintf("invalid dataset$peptides column provided to normalize_peptide_intensity_column(). Provided value: %s", paste(col_intensity, collapse = ",")), type = "error")
   }
-  if(!all(is.character(norm_algorithm)) || !any(norm_algorithm != "")) {
-    append_log(sprintf("invalid normalization algorithm provided to normalize_peptide_intensity_column(): must be (and array of) character type and contain at least one non-empty. Provided value: %s", paste(norm_algorithm, collapse = ",")), type = "error")
+  if(any(!is.character(norm_algorithm))) {
+    append_log(sprintf("invalid normalization algorithm provided to normalize_peptide_intensity_column(): must be (and array of) character type. Provided value: %s", paste(norm_algorithm, collapse = ",")), type = "error")
   }
 
   # remove NA values to ensure we remove empty rows and columns

R/plot_normalization.R

@@ -254,7 +254,8 @@ ggplot_coefficient_of_variation__leave_one_out = function(tib_input, samples, sa
   ### leave-one-out computation, per group
   tib_loo_cov = tibble()
   plotlist = list()
-  for(grp in unique(samples$group)) { #grp=unique(samples$group)[1]
+  ugrp = unique(samples$group)
+  for(grp in ugrp) { #grp=ugrp[1]
     # samples for current group
     sid = samples %>% filter(group == grp) %>% pull(sample_id)
     # we use the basic filter+normalization data as input, as this has already been pre-selected for bare minimum of detect/quant (less so than filters below) AND normalized, making CoV estimations more reflective of actual data
@@ -370,7 +371,10 @@ ggplot_coefficient_of_variation__leave_one_out = function(tib_input, samples, sa
     scale_color_manual(breaks=names(tib_plot_colors), values = tib_plot_colors, labels=names(tib_plot_colors), aesthetics = c("colour", "fill")) +
     labs(title="Effect of removing a sample prior to CoV computation on within-group CoV\nlower value = better CoV after removing sample s", y="", x="median Coefficient of Variation (%)") +
     theme_bw() +
-    theme(plot.title = element_text(size=10), legend.position = "bottom", legend.title = element_blank())
+    # alternatively, show legend when there are less than N groups. But beware that in large datasets, this causes a bug (plot+legend doesn't fit viewport)
+    # in version 1.0.9 we removed the legend because group names are also shown as rows in the plot (i.e. legend is redundant)
+    # code to reintroduce legend for small datasets only; legend.position = ifelse(length(ugrp) <= 8, "bottom", "none")
+    theme(plot.title = element_text(size=10), legend.position = "none", legend.title = element_blank())
 
   # summary stats we can print as a table, hard to glance from the figures
   tbl_loo_cov = tib_plot %>%

R/report_as_rmarkdown.R

@@ -167,7 +167,7 @@ generate_pdf_report = function(dataset, output_dir, norm_algorithm = "vwmb", rol
   # copy .Rmd file into temp directory, nested in chosen output dir
   f_newlocation = paste0(output_dir__temp, "/", basename(f))
   if(!file.copy(from = f, to = f_newlocation)) {
-    append_log(paste("failed to move report from", f, "into to the output directory:", f_newlocation), type = "error")
+    append_log(paste("failed to copy report template from", f, "into to the temp directory:", f_newlocation), type = "error")
   }
 
   ### create the actual report

R/util_generic.R

@@ -149,6 +149,7 @@ boxplot_add_text = function(bp, show_N = TRUE, show_values = T, ...) {
 #' }
 #'
 #' @param x a wide-format data.frame or tibble where the first column are assumed to represent unique row identifiers and other columns assumed 'data' columns for the output matrix
+#' @export
 as_matrix_except_first_column = function(x) {
   if (is_tibble(x)) {
     x = as.data.frame(x, stringsAsFactors = F)
@@ -168,6 +169,7 @@ as_matrix_except_first_column = function(x) {
 #' @param column_name target column name for the row names
 #' @param row_name target column name for the column names
 #' @importFrom tidyr gather
+#' @export
 matrix_to_long = function(mat, value_name = "value", column_name = "sample", row_name = "sequence") {
   x = as_tibble(mat)
   x$rownms = rownames(mat)

docker/Dockerfile

@@ -33,9 +33,9 @@ RUN R -e "devtools::install_version('remaCor', '0.0.18', upgrade = 'never', repo
 ### MS-DAP
 # from local devtools::build() file
 RUN mkdir /msdap
-COPY temp/msdap_1.0.8.tar.gz /msdap/
-RUN R -e "devtools::install_local('/msdap/msdap_1.0.8.tar.gz', upgrade = 'never')"
-# alternatively, install from github; RUN R -e "devtools::install_github('ftwkoopmans/[email protected].8', upgrade = 'never')"
+COPY temp/msdap_1.0.9.tar.gz /msdap/
+RUN R -e "devtools::install_local('/msdap/msdap_1.0.9.tar.gz', upgrade = 'never')"
+# alternatively, install from github; RUN R -e "devtools::install_github('ftwkoopmans/[email protected].9', upgrade = 'never')"
 
 
 ### finish setting up

docker/msdap_launcher_unix.sh

@@ -3,7 +3,7 @@
 # MS-DAP launch script
 # https://github.com/ftwkoopmans/msdap
 
-VERSION="1.0.8"
+VERSION="1.0.9"
 
 
 ### OS

docker/msdap_launcher_windows.ps1

@@ -1,7 +1,7 @@
 # MS-DAP launch script
 # https://github.com/ftwkoopmans/msdap
 
-$VERSION = "1.0.8"
+$VERSION = "1.0.9"
 
 
 Write-Host "$((Get-Date).ToString("HH:mm:ss")) - Starting MS-DAP launcher script for version: $VERSION"