factors_progeny_cor.Rmd

---
title: "Correlation of progeny with factors"
author: "Ricardo Ramirez"
date: "3/22/2021"
output: html_document
editor_options: 
  chunk_output_type: console
---

```{r setup, include=FALSE}
library(tidyverse)
library(Seurat)
library(corrplot)
library(ComplexHeatmap)
```

```{r}
#' Performs hierarchical clustering to a correlation matrix
#' @param cor_mat_feat_factor: correlation matrix between factors and features
get_order = function(cor_mat_feat_factor){
  cor_factors = cor(cor_mat_feat_factor)
  clust_factors = hclust(as.dist(1-cor_factors))
  
  cor_feats = cor(t(cor_mat_feat_factor))
  clust_feats = hclust(as.dist(1-cor_feats))
  
  return(list("factors" = clust_factors$labels[clust_factors$order],
              "features" = clust_feats$labels[clust_feats$order]))
}
```

```{r}
# Slide 1 is A, slide 2 is B
all_samples <- read.table(file = "./markers/NMF_Colon_combined.tsv",
                          sep = "\t",
                          header = T,
                          row.names = 1)

A1_combined_factors <- all_samples[grepl("1_1",rownames(all_samples)),]
rownames(A1_combined_factors) <- map_chr(strsplit(rownames(A1_combined_factors), split = "-"),
                                         ~ .x[[1]])

B1_combined_factors <- all_samples[grepl("1_2",rownames(all_samples)),]
rownames(B1_combined_factors) <- map_chr(strsplit(rownames(B1_combined_factors), split = "-"),
                                         ~ .x[[1]])

# Create the tibble with all useful data:
# -Factor scores, progeny scores

factor_scores <- list("V19S23-097_A1" = A1_combined_factors,
                      "V19S23-097_B1" = B1_combined_factors) %>%
  enframe(name = "slide",
          value = "f_scores") %>%
  dplyr::mutate(progeny_scores = map(slide, function(s) {
    
    slide_file <- sprintf("./results/single_slide/%s/%s.rds",
                       s,s)
    
    visium_slide <- readRDS(slide_file)
    
    PROGENy_mat <- t(as.matrix(visium_slide@assays$progeny@data))
    rownames(PROGENy_mat) <- map_chr(strsplit(rownames(PROGENy_mat), split = "-"),
                                         ~ .x[[1]])
    return(PROGENy_mat)
  }))

# Sort matrices so that they have the same barcode

factor_scores <- factor_scores %>%
  group_by(slide) %>%
  dplyr::mutate(funcomics_cor = map2(f_scores, progeny_scores, function(x, y) {
    
    combined_cor_path = cor(y[rownames(x),], x)
    
  }))
```

```{r}
# Original plots ----------------------------------------
walk(factor_scores$funcomics_cor, function(single_cor_path) {
    
    corrplot(single_cor_path[get_order(single_cor_path)$features,
                                get_order(single_cor_path)$factors],
                method = "color",is.corr = F,
                tl.col = "black",title = "individual_factors",
           col=colorRampPalette(c("darkblue","white","darkred"))(100))
    
  })
```

```{r}
# New plots 
# This is the one I defined previously
get_order = function(cor_mat_feat_factor){
  cor_factors = cor(cor_mat_feat_factor)
  clust_factors = hclust(as.dist(1-cor_factors))
  
  cor_feats = cor(t(cor_mat_feat_factor))
  clust_feats = hclust(as.dist(1-cor_feats))
  return(list("factors" = clust_factors,
              "features" = clust_feats))
}

```

# Day 0
```{r}
A1_order <- get_order(factor_scores$funcomics_cor[[1]])

A1_hmap <- Heatmap(factor_scores$funcomics_cor[[1]],
        cluster_rows = A1_order$features,
        cluster_columns = A1_order$factors,
        name = "Pearson")

draw(A1_hmap)

pdf(file = "./results/factors_annotation/hclust_A1_progeny_factors.pdf", width = 7, height = 4)

draw(A1_hmap)

dev.off()
```

# Day 14

```{r}
B1_order <- get_order(factor_scores$funcomics_cor[[2]])

B1_hmap <- Heatmap(factor_scores$funcomics_cor[[2]],
        cluster_rows = B1_order$features,
        cluster_columns = B1_order$factors,
        name = "Pearson")

draw(B1_hmap)

pdf(file = "./results/factors_annotation/hclust_B1_progeny_factors.pdf", width = 7, height = 4)

draw(B1_hmap)

dev.off()
```