Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Added mutation enrichment summary #66

Merged
merged 2 commits into from
May 2, 2022
Merged

Added mutation enrichment summary #66

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1c5e8f0
Added mutation summary plot
CamiloPosso May 2, 2022
91efaf6
Merge branch 'main' of https://github.com/CamiloPosso/BeatAMLproteomics
CamiloPosso May 2, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
24 changes: 15 additions & 9 deletions cohort_summary/NMF/multinomial_models_elastic_net/k_5_enet_multinomial_models.Rmd
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,8 @@ library(ggplot2)
library(kableExtra)
library(glmnet)
library(groupdata2)
library(gridExtra)
library(grid)

source("../cluster_model_helper.R")

Expand Down Expand Up @@ -394,32 +396,36 @@ plot_metrics <- function(chosen_alpha){
`Classification error` = rna_model_metrics$missclass_error,
data_type = "RNA"))

ggplot(summary, aes(x = Cluster, y = Recall, fill = data_type)) +
p_recall <- ggplot(summary, aes(x = Cluster, y = Recall, fill = data_type)) +
geom_bar(stat = 'identity', width = 0.33, position = position_dodge()) +
scale_fill_manual(values = data_type_colors) +
ggtitle(paste0("Recall by cluster - 5 fold cross validation, ",
ggtitle(paste0("Recall by cluster, ",
"alpha = ", chosen_alpha)) +
xlab("Cluster") + ylim(0, 1) +
theme_minimal()
ggsave(paste0("enet_multinomial_data/model_recall_enet_multinomial_", chosen_alpha, ".pdf"))

ggplot(summary, aes(x = Cluster, y = Precision, fill = data_type)) +
p_precision <- ggplot(summary, aes(x = Cluster, y = Precision, fill = data_type)) +
geom_bar(stat = 'identity', width = 0.33, position = position_dodge()) +
scale_fill_manual(values = data_type_colors) +
ggtitle(paste0("Precision by cluster - 5 fold cross validation, ",
ggtitle(paste0("Precision by cluster, ",
"alpha = ", chosen_alpha)) +
xlab("Cluster") + ylim(0, 1) +
theme_minimal()
ggsave(paste0("enet_multinomial_data/model_precision_enet_multinomial_", chosen_alpha, ".pdf"))

ggplot(summary, aes(x = Cluster, y = Classification.error, fill = data_type)) +
p_classerror <- ggplot(summary, aes(x = Cluster, y = Classification.error, fill = data_type)) +
geom_bar(stat = 'identity', width = 0.33, position = position_dodge()) +
scale_fill_manual(values = data_type_colors) +
ggtitle(paste0("Classification error by cluster - 5 fold cross validation, ",
ggtitle(paste0("Classification error by cluster, ",
"alpha = ", chosen_alpha)) +
xlab("Cluster") + ylim(0, 0.2) +
theme_minimal()
ggsave(paste0("enet_multinomial_data/model_classerror_enet_multinomial_", chosen_alpha, ".pdf"))

top_grob <- textGrob(paste0("Model metrics alpha = ", chosen_alpha),
gp = gpar(fontsize = 15))
p_all <- arrangeGrob(p_recall, p_precision, p_classerror,
ncol = 3, top = top_grob)
ggsave(paste0("enet_multinomial_data/model_metrics_alpha_", chosen_alpha, ".pdf"), p_all,
width = 15, height = 6)
}

```
Expand Down
195 changes: 194 additions & 1 deletion cohort_summary/NMF/multinomial_models_elastic_net/k_5_enet_multinomial_predictions.Rmd
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@ library(survminer)
library(survival)
library(gridExtra)
library(grid)
library(ggpattern)

source("../NMF_helper.R")
source("../cluster_model_helper.R")
Expand Down Expand Up @@ -199,7 +200,7 @@ make_prediction_plot <- function(chosen_alpha, data_type, confidence_cutoff = 0.
legend.title = element_text(size = 12, color = "black")) +
ggtitle("Predicting on 51 new samples")

p_all <- arrangeGrob(p1, p2, p3, p4, ncol = 2, nrow = 2,
p_all <- arrangeGrob(p1, p2, p4, ncol = 2, nrow = 2,
top = textGrob(paste0("Survival plots, ", data_type, ", alpha = ", chosen_alpha),
gp = gpar(fontsize = 20)))
ggsave(paste0("enet_multinomial_data/predicted_survival_",
Expand All @@ -223,4 +224,196 @@ lapply(c(0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0), function(chosen_alph



```{r}
all_mutation_data <- load_mutational_sample_data()
mutation_data <- all_mutation_data %>%
mutate(value = TRUE) %>%
pivot_wider(names_from = Gene, values_from = value, values_fn = any)
mutation_data <- mutation_data[, c("Barcode.ID", "KRAS+NRAS", "NPM1_clinical")] %>%
full_join(meta[, c("Barcode.ID", "FLT3.ITD")], by = "Barcode.ID") %>%
mutate(RAS = case_when(!is.na(`KRAS+NRAS`) ~ `KRAS+NRAS`,
TRUE ~ FALSE),
NPM1_clinical = case_when(!is.na(NPM1_clinical) ~ NPM1_clinical,
TRUE ~ FALSE),
FLT3 = case_when(FLT3.ITD == "TRUE" ~ TRUE,
TRUE ~ FALSE)) %>%
select(Barcode.ID, FLT3, NPM1_clinical, RAS)
```



```{r}
mut_enrichment <- function(sample_categories, assignments, log.scale = FALSE) {
k = length(unique(assignments$Cluster))

sample_categories <- sample_categories %>%
filter(rownames(.) %in% rownames(assignments)) %>%
left_join(assignments, by = "Barcode.ID") %>%
column_to_rownames("Barcode.ID") %>%
as.data.frame()

categories <- sample_categories %>%
select(-Cluster) %>%
colnames()

out <- lapply(categories, function(category){

p.values <- lapply(1:k, function(i){
cluster.df <- sample_categories %>%
mutate(Cluster = case_when(Cluster == i ~ TRUE,
TRUE ~ FALSE)) %>%
mutate(Cluster = as.factor(Cluster))
dat <- table(cluster.df[[category]], cluster.df$Cluster)
background_dist <- apply(dat, 1, sum)
background_prop <- background_dist[[2]]/(background_dist[[2]] + background_dist[[1]])
cluster_prop <- dat[2, 2]/(dat[1, 2] + dat[2, 2])
p_value <- fisher.test(dat, workspace = 2e8, alternative = "greater")$p.value
if (log.scale) {
p_value <- -log10(p_value)
} else {
p_value <- p_value
}
xx <- data.frame(background_prop, cluster_prop, p_value)
colnames(xx) <- c(paste("Background", category, "positive fraction"),
paste(category, "positive fraction"),
paste("Enrichment", category, "p-value"))
rownames(xx) <- paste("Cluster", i)
return(xx)
}) %>% do.call("rbind", .)

}) %>% do.call(cbind, .) %>%
as.data.frame()
return(out)
}


mut_enrichment_plot <- function(chosen_alpha, data_type, robust = TRUE){
if (robust){
path_ending <- "_robust_normalization.pdf"
} else {
path_ending <- ".pdf"
}

meta_pred <- make_prediction(chosen_alpha, data_type, robust) %>%
select(Barcode.ID, pred_cluster, pred_cluster_v2) %>%
full_join(mutation_data, by = "Barcode.ID") %>%
mutate(pred_cluster = sub("Cluster ", "", pred_cluster),
pred_cluster_v2 = sub("Cluster ", "", pred_cluster_v2))
rownames(meta_pred) <- meta_pred$Barcode.ID

sample_assn <- meta_pred[, c("Barcode.ID", "pred_cluster")] %>%
dplyr::rename(Cluster = pred_cluster) %>%
filter(!is.na(Cluster))
sample_cats <- meta_pred[rownames(sample_assn), c("Barcode.ID", "FLT3", "RAS", "NPM1_clinical")]
mutation_enrichment <- mut_enrichment(sample_cats, sample_assn) %>%
mutate(Cluster = 1:5) %>%
mutate(pred_type = "51 Samples")

sample_assn <- meta_pred[, c("Barcode.ID", "pred_cluster_v2")] %>%
dplyr::rename(Cluster = pred_cluster_v2) %>%
filter(!is.na(Cluster))
sample_cats <- meta_pred[rownames(sample_assn), c("Barcode.ID", "FLT3", "RAS", "NPM1_clinical")]
mutation_enrichment <- mut_enrichment(sample_cats, sample_assn) %>%
mutate(Cluster = 1:5) %>%
mutate(pred_type = "210 Samples") %>%
rbind(mutation_enrichment)

sample_assn <- enet_meta[, c("Barcode.ID", "k.5")] %>%
dplyr::rename(Cluster = k.5) %>%
filter(!is.na(Cluster))
sample_cats <- meta_pred[rownames(sample_assn), c("Barcode.ID", "FLT3", "RAS", "NPM1_clinical")]
mutation_enrichment <- mut_enrichment(sample_cats, sample_assn) %>%
mutate(Cluster = 1:5) %>%
mutate(pred_type = "159 Samples") %>%
rbind(mutation_enrichment) %>%
mutate(pred_type = factor(pred_type, levels = c("159 Samples",
"210 Samples",
"51 Samples")))

plot_df1 <- mutation_enrichment %>%
select(Cluster, `FLT3 positive fraction`,
`Enrichment FLT3 p-value`, pred_type) %>%
dplyr::rename(`P-value` = `Enrichment FLT3 p-value`,
`Fraction mutated` = `FLT3 positive fraction`,
`Type` = pred_type) %>%
mutate(`Significant enrichment` = case_when(`P-value` < 0.05 ~ 1,
TRUE ~ 0))
p1 <- ggplot(plot_df1, aes(x = Cluster, y = `Fraction mutated`,
group = `Type`,
fill = `Type`)) +
geom_bar(stat = "identity", pos = position_dodge(), width = 0.77, alpha = 0.7) +
scale_fill_manual(values = prediction_colors) +
geom_col_pattern(aes(x = Cluster, y = `Fraction mutated`,
pattern_alpha = `Significant enrichment`),
width = 0.77, pos = position_dodge(),
pattern = 'stripe',
color = 'black') +
scale_pattern_alpha(guide = "none") +
ggtitle("FLT3 Status") + ylim(0, 1)

plot_df2 <- mutation_enrichment %>%
select(Cluster, `RAS positive fraction`,
`Enrichment RAS p-value`, pred_type) %>%
dplyr::rename(`P-value` = `Enrichment RAS p-value`,
`Fraction mutated` = `RAS positive fraction`,
`Type` = pred_type) %>%
mutate(`Significant enrichment` = case_when(`P-value` < 0.05 ~ 1,
TRUE ~ 0))
p2 <- ggplot(plot_df2, aes(x = Cluster, y = `Fraction mutated`,
group = `Type`,
fill = `Type`)) +
geom_bar(stat = "identity", pos = position_dodge(), width = 0.77, alpha = 0.7) +
scale_fill_manual(values = prediction_colors) +
geom_col_pattern(aes(x = Cluster, y = `Fraction mutated`,
pattern_alpha = `Significant enrichment`),
width = 0.77, pos = position_dodge(),
pattern = 'stripe',
color = 'black') +
scale_pattern_alpha(guide = "none") +
ggtitle("RAS Status") + ylim(0, 1)

plot_df3 <- mutation_enrichment %>%
select(Cluster, `NPM1_clinical positive fraction`,
`Enrichment NPM1_clinical p-value`, pred_type) %>%
dplyr::rename(`P-value` = `Enrichment NPM1_clinical p-value`,
`Fraction mutated` = `NPM1_clinical positive fraction`,
`Type` = pred_type) %>%
mutate(`Significant enrichment` = case_when(`P-value` < 0.05 ~ 1,
TRUE ~ 0))
p3 <- ggplot(plot_df3, aes(x = Cluster, y = `Fraction mutated`,
group = `Type`,
fill = `Type`)) +
geom_bar(stat = "identity", pos = position_dodge(), width = 0.77, alpha = 0.7) +
scale_fill_manual(values = prediction_colors) +
geom_col_pattern(aes(x = Cluster, y = `Fraction mutated`,
pattern_alpha = `Significant enrichment`),
width = 0.77, pos = position_dodge(),
pattern = 'stripe',
color = 'black') +
scale_pattern_alpha(guide = "none") +
ggtitle("NPM1_clinical Status") + ylim(0, 1)

p_all <- arrangeGrob(p1, p2, p3, ncol = 3,
top = textGrob(paste0("Mutation Status - ", data_type, " predicions - alpha = ", chosen_alpha),
gp = gpar(fontsize = 20)))

ggsave(paste0("enet_multinomial_data/predicted_mutation_status_",
data_type, "_", chosen_alpha, path_ending),
width = 15, height = 5, p_all)

return(mutation_enrichment)
}

```



```{r}
sanity_check <- mut_enrichment_plot(0.8, "Global", TRUE)

```





44 changes: 33 additions & 11 deletions cohort_summary/NMF/multinomial_models_elastic_net/k_5_enet_multinomial_signatures.Rmd
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,8 @@ library(ggplot2)
library(glmnet)
library(msigdbr)
library(RColorBrewer)
library(gridExtra)
library(grid)

source("../cluster_model_helper.R")

Expand Down Expand Up @@ -118,7 +120,7 @@ Signature heatmap
```{r}
chosen.colors <- subtype_colors

signature_heatmap <- function(chosen_alpha, data_type){
signature_heatmap <- function(chosen_alpha, data_type, save_plot = TRUE){
if (data_type == "Global"){
data_mat <- global_mat_train
} else if (data_type == "Phospho"){
Expand Down Expand Up @@ -171,15 +173,35 @@ signature_heatmap <- function(chosen_alpha, data_type){
arrange(Cluster) %>%
select(Cluster)
sigs <- rownames(row_df)
file_path <- paste0("./enet_multinomial_data/enet_multinomial_signature_heatmap_mult_", bound_mult,
"_", data_type, "_alpha_", chosen_alpha)
file_path <- paste0("./enet_multinomial_data/signature_heatmap_",
data_type, "_alpha_", chosen_alpha)


make.pheatmap(data_mat[sigs, ], filename = file_path, format = "pdf",
scale = "none", cluster_rows = FALSE, clustering_method = "ward.D",
if (save_plot){
make.pheatmap(data_mat[sigs, ], filename = file_path, format = "pdf",
scale = "none", cluster_rows = FALSE, clustering_method = "ward.D",
annotation_col = col_df, annotation_row = row_df, annotation_colors = ann_colors,
breaks = heatmap_breaks, color = heatmap_colors, show_rownames = F, show_colnames = F,
main = paste0(length(sigs), " total features, alpha = ", chosen_alpha, " ", data_type),
height = 7, width = 7)
}

return(pheatmap(data_mat[sigs, ], scale = "none", cluster_rows = FALSE, clustering_method = "ward.D",
annotation_col = col_df, annotation_row = row_df, annotation_colors = ann_colors,
breaks = heatmap_breaks, color = heatmap_colors, show_rownames = F, show_colnames = F,
main = paste0(length(sigs), " total features, alpha = ", chosen_alpha, " ", data_type), height = 7, width = 7)
main = paste0(length(sigs), " total features, alpha = ", chosen_alpha, " ", data_type)))
}

signature_heatmap_v2 <- function(chosen_alpha){
p_global <- signature_heatmap(chosen_alpha, "Global", save_plot = FALSE)
p_phospho <- signature_heatmap(chosen_alpha, "Phospho", save_plot = FALSE)
p_RNA <- signature_heatmap(chosen_alpha, "RNA", save_plot = FALSE)

top_grob <- textGrob(paste0("Signatures alpha = ", chosen_alpha),
gp = gpar(fontsize = 20))
p_all <- arrangeGrob(p_global[[4]], p_RNA[[4]], p_phospho[[4]],
ncol = 3, top = top_grob)
ggsave(paste0("enet_multinomial_data/signature_heatmaps_alpha_", chosen_alpha, ".pdf"), p_all,
width = 30, height = 12)
}

```
Expand All @@ -188,12 +210,12 @@ signature_heatmap <- function(chosen_alpha, data_type){

```{r}
lapply(c(0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0), function(chosen_alpha){
signature_heatmap(chosen_alpha, "Global")
signature_heatmap(chosen_alpha, "Phospho")
signature_heatmap(chosen_alpha, "RNA")
signature_heatmap_v2(chosen_alpha)
})


# signature_heatmap(0.7, "Global")
# signature_heatmap(0.7, "RNA")
# signature_heatmap(0.7, "Phospho")

```

Expand Down
Loading