heatmaps-4libraries.Rmd

---
title: "heatmaps-4libraries"
output: html_document
---
Rmd to make heatmaps using `pheatmap` from the lists of DEGs from the Rmd: "DESeq2.md". 

Load in packages:
```{r}
library(tidyverse)
library(pheatmap)
library(RColorBrewer)
library(data.table)
```

### Read in DEG lists:

List of 408 DEGs (infection comparison taking temperature into account):  
```{r}
degs_inftemp <- read.delim("../analyses/DESeq2/DEGlist-infection-with-temp-counts.tab", sep = '\t')
head(degs_inftemp)
```

List of 123 DEGs (infectin DEGs influenced by temperature treatment):
```{r}
degs_contrast.temp <- read.delim("../analyses/DESeq2/DEGlist-contrast_temperature-counts.tab", sep = '\t')
head(degs_contrast.temp)
```

# Make heatmaps of the two lists of count data

Set heatmap colors:
```{r}
heatmapPurpleOrange <- RColorBrewer::brewer.pal(11, "PuOr")
```

## Heatmap of counts from list of 408 degs   
infection comparison taking temperature into account

Set Trinity_ID column as rownames:
```{r}
rownames(degs_inftemp) <- degs_inftemp$Trinity_ID #set Trinity IDs as rownames
inftemp.heatmap <- degs_inftemp[,-1] #remove redundant column
head(inftemp.heatmap)
```

Pull out just the sample count data:
```{r}
inftemp.heatmap <- select(inftemp.heatmap, "X380822_cold_uninfected", "X380823_cold_infected", "X380824_warm_uninfected", "X380825_warm_infected")
head(inftemp.heatmap)
```

rename columns for easier viewing in heatmap:
```{r}
colnames(inftemp.heatmap) <- c("380822", "380823", "380824", "380825")
head(inftemp.heatmap)
```

Going to remove all rows that have "0" for all samples: 
```{r}
noz_inftemp.heatmap <- inftemp.heatmap[rowSums(inftemp.heatmap[, -1] >0) !=0, ]
head(noz_inftemp.heatmap)
```

```{r}
pheatmap(noz_inftemp.heatmap, scale = "row", cluster_row = TRUE, cluster_cols = FALSE, clustering_distance_rows = "euclidean", clustering_method = "average", show_rownames = FALSE, show_colnames = TRUE, treeheight_col = 80, legend = TRUE, color = heatmapPurpleOrange, fontsize_col = 15, fontsize_row = 12)
```
408 DEGs. Infection DEGs taking temperature into account. 

### `join` with `blast` output and uniprot-SP-GO:

Load in `blast` output and uniprot-SP-GO file:
```{r}
blast_GO <- read.delim("../analyses/BLAST-to-GOslim/transcriptomev3.1-blast-uniprotGO.tab", header = TRUE)
head(blast_GO)
```


Set rownames for inftemp.heatmap as a column called "Trinity_ID":
```{r}
inftemp.heatmap <- tibble::rownames_to_column(inftemp.heatmap, "Trinity_ID")
head(inftemp.heatmap)
```

`join` the previous file with inftemp.heatmap:
```{r}
blast_uniprot_inftemp <- left_join(inftemp.heatmap, blast_GO, by = "Trinity_ID")
head(blast_uniprot_inftemp)
```

write out table with counts, blast output, and uniprot-SP-GO:
```{r}
#write.table(blast_uniprot_inftemp, "../analyses/pheatmap/infection-temp-DEGs-counts_annot.tab", sep = '\t', quote = FALSE, row.names = FALSE)
```
Wrote out 02/11/2021. Commented out code. 

----

## Heatmap of counts from list of 123 degs 
infection DEGs influenced by temperature treatment    
USED in PAPER

Set Trinity_ID as row names:
```{r}
rownames(degs_contrast.temp) <- degs_contrast.temp$Trinity_ID #set Trinity IDs as rownames
contrasttemp.heatmap <- degs_contrast.temp[,-1] #remove redundant column
head(contrasttemp.heatmap)
```

Pull out just the sample count data:
```{r}
contrasttemp.heatmap <- select(contrasttemp.heatmap, "X380822_cold_uninfected", "X380823_cold_infected", "X380824_warm_uninfected", "X380825_warm_infected")
head(contrasttemp.heatmap)
```

rename columns for easier viewing in heatmap:
```{r}
colnames(contrasttemp.heatmap) <- c("380822", "380823", "380824", "380825")
head(contrasttemp.heatmap)
```

Going to remove all rows that have "0" for all samples: 
```{r}
noz_contrasttemp.heatmap <- contrasttemp.heatmap[rowSums(contrasttemp.heatmap[, -1] >0) !=0, ]
head(noz_contrasttemp.heatmap)
```

```{r}
pheatmap(noz_contrasttemp.heatmap, scale = "row", cluster_row = TRUE, cluster_cols = FALSE, clustering_distance_rows = "euclidean", clustering_method = "average", show_rownames = FALSE, show_colnames = TRUE, treeheight_col = 80, legend = TRUE, color = heatmapPurpleOrange, fontsize_col = 12, fontsize_row = 12)
```

### Annotate heatmap #2 with color-coded cluster assignments based on the cladogram (will be used in paper)

Set heatmap for temperature contrast as an object:
```{r}
results <- pheatmap(noz_contrasttemp.heatmap, scale = "row", cluster_row = TRUE, cluster_cols = FALSE, clustering_distance_rows = "euclidean", clustering_method = "average", show_rownames = FALSE, show_colnames = TRUE, treeheight_col = 80, legend = TRUE, color = heatmapPurpleOrange, fontsize_col = 12, fontsize_row = 12)
```

Extract 4 clusters:
```{r}
contrast.clust <- cbind(noz_contrasttemp.heatmap,
                   cluster = cutree(results$tree_row,
                                     k = 4))
head(contrast.clust)
```

Pull out just the column of cluster numbers:
```{r}
clusters <- select(contrast.clust, "cluster")
head(clusters)
```

Rename the clusters in the rows to "cluster #" instead of just the number. 
```{r}
clusters <- data.frame(cluster = ifelse(clusters == 1, "cluster1", ifelse(clusters == 2, "cluster2", ifelse(clusters == 3, "cluster3", "cluster4"))))
head(clusters)
```

Add column annotations for the samples denoting temperature treatment and infection status:
```{r}
sample_col <- data.frame(temperature = c("decreased", "decreased", "elevated", "elevated"))
row.names(sample_col) <-colnames(noz_contrasttemp.heatmap)
sample_col
```

combine the two sample info into one dataframe:
```{r}
sample_col$infection.status <- c("uninfected", "infected", "uninfected", "infected")
sample_col
```

specify colors for annotation columns:
```{r}
my_colors <- list(
  temperature = c(decreased = "blue", elevated = "red"),
  infection.status = c(infected = "darkgreen", uninfected = "yellow"),
  cluster = c(cluster1 = "#1B9E77", cluster2 = "#D95F02", cluster3 = "#7570B3", cluster4 = "#E7298A")
)
```

```{r}
pheatmap(noz_contrasttemp.heatmap, scale = "row", cluster_row = TRUE, cluster_cols = FALSE, clustering_distance_rows = "euclidean", clustering_method = "average", annotation_colors = my_colors, annotation_row = clusters, annotation_col = sample_col, show_rownames = FALSE, show_colnames = TRUE, treeheight_col = 80, legend = TRUE, color = heatmapPurpleOrange, fontsize_col = 12, fontsize_row = 12)
```
Saved heatmap by right clicking > Save Image As, download as png, then open in preview, export > PDF, save. 


save heatmap to analyses/pheatmap:
```{r}
#pdf("../analyses/pheatmap/tempinfluenced-infectionDEGs-heatmap.pdf", width = 11, height = 8.5)
#pheatmap(noz_contrasttemp.heatmap, scale = "row", cluster_row = TRUE, cluster_cols = FALSE, clustering_distance_rows = "euclidean", clustering_method = "average", annotation_colors = my_colors, annotation_row = clusters, annotation_col = sample_col, show_rownames = FALSE, show_colnames = TRUE, treeheight_col = 80, legend = TRUE, color = heatmapPurpleOrange, fontsize_col = 12, fontsize_row = 12)
#dev.off()
```
saved earlier. commented out code 08/26/2020. 


### `join` count data from heatmap with transcriptome v 3.1 `blast` output and uniprot-SP-GO annotation:


Set rownames in heatmap matrix as column called "Trinity_ID":

```{r}
contrast.clust <- tibble::rownames_to_column(contrast.clust, "Trinity_ID")
head(contrast.clust)
```


`join` the transcriptome v 3.1 blast output and uniprot GO file with contrast.clust:
```{r}
blast_uniprot_contrast <- left_join(contrast.clust, blast_GO, by = "Trinity_ID")
head(blast_uniprot_contrast)
```

write out table:
```{r}
#write.table(blast_uniprot_contrast, "../analyses/pheatmap/contrasttemp_DEGs-clusters-annot.tab", sep = '\t', quote = FALSE, row.names = FALSE)
```
Wrote out 02/11/2021. Commented out code.