improving documentation

.Rbuildignore

@@ -6,3 +6,4 @@
 ^pkgdown$
 ^codecov\.yml$
 ^\.github$
+^vignettes/articles$

DESCRIPTION

@@ -5,6 +5,7 @@ Version: 0.1.0
 Author: Azad Sadr
 Maintainer: Azad Sadr <[email protected]>
 Authors@R: c(
+    person("Elena", "Buscaroli", role=c("aut", "cre"), email="[email protected]"),
     person("Azad", "Sadr", role=c("aut", "cre"), email="[email protected]"),
     person("Giulio", "Caravagna", role=c("aut"), email="[email protected]")
     )

NAMESPACE

@@ -5,16 +5,21 @@ S3method(print,basilica_obj)
 export(configure_environment)
 export(fit)
 export(fit_clustering)
+export(get_denovo_signames)
+export(get_fixed_signames)
 export(get_input)
+export(get_signames)
 export(have_loaded_env)
 export(have_python_deps)
 export(load_conda_env)
 export(merge_clusters)
 export(plot_data)
 export(plot_exposures)
+export(plot_scores)
 export(plot_signatures)
 export(plot_similarity_reference)
 export(print.basilica_obj)
+export(refine_denovo_signatures)
 export(which_conda_env)
 import(ggplot2)
 importFrom(cli,cli_abort)

R/getters.R

@@ -1,17 +1,39 @@
 # Signatures #####
 ## names ####
+
+#' Get signatures names
+#'
+#' @param x basilica object.
+#' @param types List of variant types to retrieve signames for.
+#'
+#' @return list of signature names with names equal to `types`.
+#' @export
 get_signames = function(x, types=get_types(x)) {
   lapply(types, function(t)
     get_signatures(x, types=t, matrix=TRUE)[[t]] %>% rownames()) %>%
     setNames(types)
 }
 
+#' Get reference signatures names
+#'
+#' @param x basilica object.
+#' @param types List of variant types to retrieve signames for.
+#'
+#' @return list of reference signature names with names equal to `types`.
+#' @export
 get_fixed_signames = function(x, types=get_types(x)) {
   lapply(types, function(t)
     get_fixed_signatures(x, types=t, matrix=TRUE)[[t]] %>% rownames()) %>%
     setNames(types)
 }
 
+#' Get denovo signatures names
+#'
+#' @param x basilica object.
+#' @param types List of variant types to retrieve signames for.
+#'
+#' @return list of de novo signature names with names equal to `types`.
+#' @export
 get_denovo_signames = function(x, types=get_types(x)) {
   lapply(types, function(t)
     get_denovo_signatures(x, types=t, matrix=TRUE)[[t]] %>% rownames()) %>%

R/plot_QC.R

@@ -35,16 +35,30 @@ plot_gradient_norms = function(x, types=get_types(x)) {
 
 
 
-plot_scores = function(x, types=get_types(x)) {
+#' Plot model selection scores
+#'
+#' @description
+#' Function to plot the model selection scores for NMF on each variant type and clustering.
+#' Reported scores are BIC and negative log-likelihood.
+#'
+#' @param x basilica object.
+#' @param types List of variant types to visualize.
+#' @param remove_outliers Logical. If `TRUE`, outliers in each score will be removed.
+#'
+#' @return ggplot2 object.
+#' @export
+plot_scores = function(x, types=get_types(x), remove_outliers=FALSE) {
   scores = get_scores(x) %>%
     dplyr::group_by(score_id, parname, type) %>%
     dplyr::mutate(is.min=score==min(score),
                   label=replace(is.min, is.min==T, "Best fit"),
-                  label=replace(label, label==F | score_id=="likelihood", NA)) %>%
+                  label=replace(label, label==F | score_id=="likelihood", NA))
 
-    dplyr::group_by(score_id, parname, type) %>%
-    dplyr::mutate(is.outlier=score %in% boxplot.stats(score)$out) %>%
-    dplyr::filter(!is.outlier)
+  if (remove_outliers)
+    scores = scores %>%
+      dplyr::group_by(score_id, parname, type) %>%
+      dplyr::mutate(is.outlier=score %in% boxplot.stats(score)$out) %>%
+      dplyr::filter(!is.outlier)
 
   scores_nmf = scores %>%
     dplyr::filter(parname == "K") %>%

R/utils_refine.R

@@ -1,3 +1,17 @@
+#' De novo signatures refinement.
+#'
+#' @description
+#' Function to refine the inferred de novo signatures.
+#' The function performs a linear combination between de novo and reference
+#' signatures. If a de novo signature can be explained as a linear combination
+#' of one or more reference signatures, it will be removed and its exposures
+#' will be distributed among the similar signatures.
+#'
+#' @param x basilica object.
+#' @param types List of variant types to perform de novo refinement on.
+#'
+#' @return basilica object.
+#' @export
 refine_denovo_signatures = function(x, types=get_types(x)) {
 
   alternatives = get_alternatives(x) %>% dplyr::filter(type %in% types)

_pkgdown.yml

@@ -12,11 +12,15 @@ reference:
   contents:
   - fit
   - fit_clustering
+  - refine_denovo_signatures
 
 - title: Getters
   desc:  Utility functions to extract attributes.
   contents:
   - get_input
+  - get_signames
+  - get_fixed_signames
+  - get_denovo_signames
 
 - title: Plotting functions
   desc:  Plotting functions available in the package
@@ -27,6 +31,7 @@ reference:
   - merge_clusters
   - plot_fit
   - plot_similarity_reference
+  - plot_scores
 
 - title: S3 objects
   desc:  Print and plot S3 functions

nobuild/test_basilica.R

@@ -2,6 +2,8 @@ devtools::load_all("~/GitHub/simbasilica/")
 load_deps()
 devtools::load_all("~/GitHub/basilica/")
 
+
+## SIMULATED ####
 counts = get_input(example_dataset, matrix=T)
 reference_cat = list("SBS"=COSMIC_sbs_filt, "DBS"=COSMIC_dbs)
 max_K = sapply(get_signames(example_dataset), length) %>% max()
@@ -17,3 +19,34 @@ x_refined_cluster %>% merge_clusters() %>% plot_exposures()
 example_dataset %>% plot_signatures()
 x_refined_cluster %>% merge_clusters() %>% plot_signatures()
 
+
+
+
+
+
+## REAL ####
+x = readRDS("~/Google Drive/My Drive/work/basilica_shared/fit_27052024/fit_wcat.Breast.Rds")
+for (tid in get_types(x)) {
+  alt_t = x$nmf[[tid]]$pyro$alternatives
+
+  x$nmf[[tid]]$pyro$alternatives = alt_t %>%
+    dplyr::rowwise() %>%
+    dplyr::mutate(seed=pyro_fit[[1]]$seed,
+                  pyro_fit=list(pyro_fit[[1]])) %>%
+    dplyr::ungroup()
+}
+
+x$clustering$pyro$alternatives = x$clustering$pyro$alternatives %>%
+  dplyr::rowwise() %>%
+  dplyr::mutate(seed=pyro_fit[[1]]$pyro$seed,
+                pyro_fit=list(pyro_fit[[1]])) %>%
+  dplyr::ungroup()
+
+x_refined = refine_denovo_signatures(x, types="SBS")
+
+saveRDS(x_refined, "~/Desktop/fit_wcat.Breast_refined.Rds")
+
+x %>% plot_signatures()
+
+cls_list = x %>% plot_cluster_scores()
+x %>% plot_cls_score_heatmap()

vignettes/.gitignore

@@ -0,0 +1,2 @@
+*.html
+*.R