1.1_timepointSpecificGenes.Rmd

---
title: "RNA-seq Mouse Placenta"
author: "Ha T. H. Vu"
output: html_document
---
  
```{r setup, include=FALSE}
options(max.print = "75")
knitr::opts_chunk$set(
  echo = TRUE,
  collapse = TRUE,
  comment = "#>",
  fig.path = "Files/",
  fig.width = 8,
  prompt = FALSE,
  tidy = FALSE,
  message = FALSE,
  warning = TRUE
)
knitr::opts_knit$set(width = 75)
```

Documentation for the analysis of mouse placental RNA-seq data at e7.5, e8.5 and e9.5.

## 1. Hierarchical clustering analysis
We cluster transcripts with hierarchical clustering. Transcripts with low raw counts (mean raw counts < 20 in more than one time point) were filtered out, and expression data (in TPM) was scaled and re-centered. Hierarchical clustering was performed in the top 75% most variable protein coding transcripts (23,571 transcripts total). <br>
First, we filter transcripts based on raw counts.
```{r}
library("dplyr", suppressMessages())
library("tidyverse", suppressMessages())
library("dendextend", suppressMessages())
library("matrixStats", suppressMessages())

###building transcript names =====
t2g <- read.table("Files/t2g.txt", header = T, sep = "\t")
t2g <- t2g[order(t2g$target_id),]

#load protein-coding transcripts
coding <- read.table("Files/Mus_musculus_grcm38_coding_transcripts.txt", header = F)

###filter based on raw est. counts and remove outliers =====
est_counts <- read.csv("Files/estCounts_allSamples.tsv", header = TRUE, sep = "\t", stringsAsFactors=FALSE)

#reformat count file
est_counts <- est_counts[order(est_counts$target_id),]
rownames(est_counts) <- est_counts$target_id #make transcript names to be row names
est_counts <- est_counts[, -c(1, 20, 21)] #exclude non-count columns
est_counts <- est_counts[, -which(names(est_counts) %in% c("E7.5_1", "E9.5_6"))] #remove outliers

mean_cts <- data.frame(trans=rownames(est_counts),
                       E7.5_mean=NA,
                       E8.5_mean=NA,
                       E9.5_mean=NA)
mean_cts$E7.5_mean <- rowMeans(est_counts[, c("E7.5_2", "E7.5_3", "E7.5_4", "E7.5_5", "E7.5_6")])
mean_cts$E8.5_mean <- rowMeans(est_counts[, c("E8.5_1", "E8.5_2", "E8.5_3", "E8.5_4", "E8.5_5", "E8.5_6")])
mean_cts$E9.5_mean <- rowMeans(est_counts[, c("E9.5_1", "E9.5_2", "E9.5_3", "E9.5_4", "E9.5_5")])
row.names(mean_cts) <- mean_cts$trans
mean_cts <- mean_cts[,2:ncol(mean_cts)]

#filter low count transcripts
basic_filter <- function (row, min_reads = 20, min_prop = 1/3) {
  mean(row >= min_reads) >= min_prop
}
keep <- apply(mean_cts, 1, basic_filter)
```
Next, transform TPM values of transcripts for clustering.
```{r}
###load TPM file in to do clustering =====
tpm <- read.csv("Files/TPM_allSamples.tsv", header = TRUE, sep = "\t", stringsAsFactors=FALSE)
tpm <- tpm[order(tpm$target_id),]
row.names(tpm) <- tpm$target_id
tpm <- tpm[,-c(1, 20, 21)]
tpm <- tpm[, -which(names(tpm) %in% c("E7.5_1", "E9.5_6"))]
tpm[] <- lapply(tpm, function(x) as.numeric(as.character(x)))
dim(tpm)[1] #number of transcripts originally

#filter out the low count transcripts according to the est_counts filtering above
tpm <- tpm[keep, ]
dim(tpm)[1] #number of transcripts after basic filtering

mean_tpm <- data.frame(trans=rownames(tpm),
                       E7.5_mean=NA,
                       E8.5_mean=NA,
                       E9.5_mean=NA)
mean_tpm$E7.5_mean <- rowMeans(tpm[, c("E7.5_2", "E7.5_3", "E7.5_4", "E7.5_5", "E7.5_6")])
mean_tpm$E8.5_mean <- rowMeans(tpm[, c("E8.5_1", "E8.5_2", "E8.5_3", "E8.5_4", "E8.5_5", "E8.5_6")])
mean_tpm$E9.5_mean <- rowMeans(tpm[, c("E9.5_1", "E9.5_2", "E9.5_3", "E9.5_4", "E9.5_5")])
row.names(mean_tpm) <- mean_tpm$trans
mean_tpm <- mean_tpm[,2:ncol(mean_tpm)]
mean_tpm$var <- rowVars(as.matrix(mean_tpm)) #calculate the variance
mean_tpm_keep <- subset(mean_tpm, mean_tpm$var > quantile(mean_tpm$var, 0.25)) #keep the top 75% most variable transcripts

tpm1 <- subset(tpm, rownames(tpm) %in% rownames(mean_tpm_keep)) #keep the top 75% most variable transcripts
dim(tpm1)[1]
tpm1 <- tpm1[which(row.names(tpm1) %in% coding$V1),] #keep only coding transcripts
tpm1[] <- lapply(tpm1, function(x) as.numeric(as.character(x)))

#center and scale the data, for further analysis
tpm2 <- apply(tpm1, 1, function(x) (x-mean(x))/sd(x))
tpm2 <- t(tpm2)
tpm2 <- as.data.frame(tpm2)
dim(tpm2)[1] #number of transcripts for downstream
#write.table(tpm2, "Files/tpmForClustering.txt", quote = F, row.names = T, sep = "\t")
```
After filtering and getting 23571 protein-coding transcripts, we carry out hierarchical clustering.
```{r}
###hierarchical clustering =====
hc <- hclust(dist(tpm2, method = "euclidean"), "complete")
#save(hc, file = "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2B_hierarchicalClustering/hierClust.rda")
dend <- as.dendrogram(hc)
#save(dend, file = "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2B_hierarchicalClustering/dendrogram.rda")
```
To identify groups, we cut the tree at the highest level and obtain 3 groups. The number of transcripts in each group is: <br>
- Group e7.5: 8242 transcripts (= 5566 genes) <br>
- Group e8.5: 8091 transcripts (= 5536 genes) <br>
- Group e9.5: 7238 transcripts (= 5347 genes) <br>
```{r}
#cutree to make clusters
trans_cluster <- cutree(hc, k = 3) %>% enframe()
table(trans_cluster$value) #number of transcripts in each group
```
Next, we summarize the transcripts per group with mean scaled TPM for plotting purposes.
```{r, eval = F}
###summarize the time points by mean scaled tpm ====
tpm3 <- tpm2
tpm3 <- cbind(rownames(tpm2), tpm3)
rownames(tpm3) <- 1:nrow(tpm3)
colnames(tpm3)[1] <- "transcripts"

E7.5_mean <- rowMeans(tpm3[,c("E7.5_2", "E7.5_3", "E7.5_4", "E7.5_5", "E7.5_6")])
E8.5_mean <- rowMeans(tpm3[,c("E8.5_1", "E8.5_2", "E8.5_3", "E8.5_4", "E8.5_5", "E8.5_6")])
E9.5_mean <- rowMeans(tpm3[,c("E9.5_1", "E9.5_2", "E9.5_3", "E9.5_4", "E9.5_5")])

#e7.5
e7.5_table <- data.frame(tpm3$transcripts, E7.5_mean)
e7.5_table$time <- c("e7.5")
rownames(e7.5_table) <- 1:nrow(e7.5_table)
colnames(e7.5_table) <- c("transcripts", "mean_cts_scaled", "time")

#e8.5
e8.5_table <- data.frame(tpm3$transcripts, E8.5_mean)
e8.5_table$time <- c("e8.5")
rownames(e8.5_table) <- 1:nrow(e8.5_table)
colnames(e8.5_table) <- c("transcripts", "mean_cts_scaled", "time")

#e9.5
e9.5_table <- data.frame(tpm3$transcripts, E9.5_mean)
e9.5_table$time <- c("e9.5")
rownames(e9.5_table) <- 1:nrow(e9.5_table)
colnames(e9.5_table) <- c("transcripts", "mean_cts_scaled", "time")

summary <- rbind(rbind(e7.5_table, e8.5_table), e9.5_table)
summary <- summary[order(summary$transcripts),]
row.names(summary) <- 1:nrow(summary)
summary <- inner_join(summary, trans_cluster, by = c("transcripts" = "name"))
summary <- inner_join(summary, t2g, by = c("transcripts" = "target_id"))
#write.table(summary, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2B_hierarchicalClustering/transcriptGroups.txt", sep  = "\t", row.names = F, quote = F)

#group 1
geneGroup1 <- subset(summary, summary$value == "1")
length(unique(geneGroup1$transcripts))
length(unique(geneGroup1$ens_gene))
#write.table(geneGroup1, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2B_hierarchicalClustering/e8.5Group.txt", sep  = " ", quote = F)

#group 2
geneGroup2 <- subset(summary, summary$value == "2")
length(unique(geneGroup2$transcripts))
length(unique(geneGroup2$ens_gene))
#write.table(geneGroup2, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2B_hierarchicalClustering/e9.5Group.txt", sep  = " ", quote = F)

#group 3
geneGroup3 <- subset(summary, summary$value == "3")
length(unique(geneGroup2$transcripts))
length(unique(geneGroup3$ens_gene))
#write.table(geneGroup3, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2B_hierarchicalClustering/e7.5Group.txt", sep  = " ", quote = F)
```

## 2. Differential expression analysis:
We carry out differential expression analysis across the three timepoints and identified transcripts and genes with the strongest changes over time. It is possible to run this analysis with a personal laptop, but I'd recommend running it on HPC. <br>
We set up the analysis by loading some necessary files.
```{r, eval = F}
library("sleuth")
library("dplyr")

set.seed(123)

###building sample description
dir <- "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/1_kallisto/official/"
sample_id <- dir(file.path(dir))
sample_id <- sample_id[grep("^S", sample_id)] #remove the unrelated file if there is

kal_dirs <- file.path(dir, sample_id, "abundance.h5")

desc <- read.table("Files/0.0_sampleNameMapping.csv", sep = "\t", header = F)
desc$condition <- substr(desc$V1, 1, 4)
desc <- desc[order(desc$V2),]
desc <- dplyr::mutate(desc, path = kal_dirs)
desc <- desc[,2:ncol(desc)]
colnames(desc) <- c("sample", "condition", "path")
desc <- desc[order(desc$condition), ]
desc <- subset(desc, !(desc$ID %in% c("S55", "S56"))) #remove sample S55, S56 outlier

e7.5vsE8.5 <- subset(desc, (desc$condition %in% c("E7.5", "E8.5")))
e8.5vsE9.5 <- subset(desc, (desc$condition %in% c("E9.5", "E8.5")))
e7.5vsE9.5 <- subset(desc, (desc$condition %in% c("E7.5", "E9.5")))

###building transcript name table. Since the object is presaved, just load it here
t2g <- read.table("Files/t2g.txt", header = T, sep = "\t")
#load protein-coding transcripts
coding <- read.table("Files/Mus_musculus_grcm38_coding_transcripts.txt", header = F)
```
Build functions to run DEG analysis. The `deg()` function below is one which uses likelihood ratio test in Sleuth to identify DE genes and transcripts. It will return a result table (named `deg`) with DE protein-coding transcripts that belong to DE genes, and a table of all transcripts + necessary metrics (named `trans`), for plotting.
```{r, eval = F}
deg <-  function(description, targetMap, con1, con2){
  #### 1.1. gene level
  so <- sleuth_prep(description, target_mapping = targetMap, aggregation_column = "ens_gene", extra_bootstrap_summary = TRUE)
  so <- sleuth_fit(so, ~condition, 'full')
  so <- sleuth_fit(so, ~1, 'reduced')
  so <- sleuth_lrt(so, 'reduced', 'full')
  
  sleuth_table <- sleuth_results(so, 'reduced:full', 'lrt', show_all = FALSE, pval_aggregate = so$pval_aggregate)
  sleuth_table <- dplyr::filter(sleuth_table, qval <= 0.05)
  
  #### 1.2. transcript level
  soTrans <- sleuth_prep(description, target_mapping = targetMap, extra_bootstrap_summary = TRUE)
  soTrans <- sleuth_fit(soTrans, ~condition, 'full')
  soTrans <- sleuth_fit(soTrans, ~1, 'reduced')
  soTrans <- sleuth_lrt(soTrans, 'reduced', 'full')
  
  sleuth_tableTrans <- sleuth_results(soTrans, 'reduced:full', 'lrt', show_all = FALSE, pval_aggregate = soTrans$pval_aggregate)
  
  ## finding up/down regulated transcripts
  cond1 <- description$sample[description$condition == con1]
  cond2 <- description$sample[description$condition == con2]
  
  raw_cond1 <- subset(soTrans$obs_raw, soTrans$obs_raw$sample %in% cond1)
  raw_cond2 <- subset(soTrans$obs_raw, soTrans$obs_raw$sample %in% cond2)
  
  #calculate mean TPM of each transcript across all samples
  means_cond1 <- aggregate(tpm~target_id, data = raw_cond1, FUN=function(x) c(mean=mean(x)))
  colnames(means_cond1) <- c("target_id", "TPM_con1")
  
  means_cond2 <- aggregate(tpm~target_id, data = raw_cond2, FUN=function(x) c(mean=mean(x)))
  colnames(means_cond2) <- c("target_id", "TPM_con2")
  
  merged_means <- merge(means_cond1, means_cond2, by = c("target_id"))
  
  sleuth_tableTrans_TPM <- left_join(sleuth_tableTrans, merged_means)
  
  #calculate FC between 2 condition
  sleuth_tableTrans_TPM$TPM_con1 <- sleuth_tableTrans_TPM$TPM_con1 + 1
  sleuth_tableTrans_TPM$TPM_con2 <- sleuth_tableTrans_TPM$TPM_con2 + 1
  sleuth_tableTrans_TPM <- transform(sleuth_tableTrans_TPM, log2FC = log2(sleuth_tableTrans_TPM$TPM_con2 / sleuth_tableTrans_TPM$TPM_con1))
  
  #keep only DE transcripts of DE genes
  sleuth_tableTrans_TPM <- subset(sleuth_tableTrans_TPM,
                                  sleuth_tableTrans_TPM$ens_gene %in% sleuth_table$target_id
                                  & sleuth_tableTrans_TPM$qval <= 0.05
                                  & abs(sleuth_tableTrans_TPM$log2FC) >= log2(1.5))
  sleuth_tableTrans_TPM <- sleuth_tableTrans_TPM[order(sleuth_tableTrans_TPM$ext_gene),]
  
  #filter out non coding transcripts
  sleuth_tableTrans_TPM <- sleuth_tableTrans_TPM[sleuth_tableTrans_TPM$target_id %in% coding$V1,]
  sleuth_tableTrans <- sleuth_tableTrans[sleuth_tableTrans$target_id %in% coding$V1,]
  
  
  results <- list("deg" = sleuth_tableTrans_TPM, "trans" = sleuth_tableTrans)
  return(results)
}
```
Use the `deg()` function to analyze e7.5 vs e8.5, e8.5 vs e9.5, and e7.5 vs e9.5. Notes: There may be some warnings like this:
"NA values were found during variance shrinkage estimation due to mean observation values outside of the range used for the LOESS fit". Since the number of NA transcripts is small (1 to 6 transcripts), and the NA values won't affect the adjustment of p-values, we may ignore the warnings. <br>
For **e7.5 vs e8.5** test, run the following code. In the end, we obtain: <br>
+ Total number of DE transcripts: 2703 <br>
+ Total number of DE genes: 2325 (this number may not be equal genes-up-at-E7.5 + genes-up-at-E8.5 since one gene can be up-regulated at both timepoints if it has different transcripts up-regulated at these timepoints.) <br>
+ Number of transcripts up at e7.5: 740 <br>
+ Number of genes up at e7.5: 652 <br>
+ Number of transcripts up at e8.5: 1963 <br>
+ Number of genes up at e8.5: 1700 <br>
```{r, eval = F}
# e7.5 vs e8.5 test
e7.5_e8.5 <- deg(e7.5vsE8.5, t2g, "e7.5", "e8.5")

e7.5vsE8.5deg <- e7.5_e8.5$deg
length(unique(e7.5vsE8.5deg$target_id)) #check total number of DE transcripts
length(unique(e7.5vsE8.5deg$ens_gene)) #check total number of DE genes

e7.5_e7.5vsE8.5 <- e7.5vsE8.5deg[e7.5vsE8.5deg$log2FC <= -log2(1.5),]
length(unique(e7.5_e7.5vsE8.5$target_id)) #check total number of transcripts up-regulated at e7.5, compared to e8.5
length(unique(e7.5_e7.5vsE8.5$ens_gene)) #check total number of genes up-regulated at e7.5, compared to e8.5
#write.table(e7.5_e7.5vsE8.5, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2A_sleuth/e7.5_e7.5vsE8.5_DEtransGenes.txt", row.names = F, quote = F, sep = "\t")

e8.5_e7.5vsE8.5 <- e7.5vsE8.5deg[e7.5vsE8.5deg$log2FC >= log2(1.5),]
length(unique(e8.5_e7.5vsE8.5$target_id)) #check total number of transcripts up-regulated at e8.5, compared to e7.5
length(unique(e8.5_e7.5vsE8.5$ens_gene)) #check total number of genes up-regulated at e8.5, compared to e7.5
#write.table(e8.5_e7.5vsE8.5, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2A_sleuth/e8.5_e7.5vsE8.5_DEtransGenes.txt", row.names = F, quote = F, sep = "\t")
```
For **e7.5 vs e9.5** test: <br>
+ Number of DE transcripts: 3948 <br>
+ Number of DE genes: 3190 <br>
+ Number of transcripts up at e7.5: 1154 <br>
+ Number of genes up at e7.5: 935 <br>
+ Number of transcripts up at e9.5: 2794 <br>
+ Number of genes up at e9.5: 2264 <br>
```{r, eval=F}
# e7.5 vs e9.5 test
e7.5_e9.5 <- deg(e7.5vsE9.5, t2g, "e7.5", "e9.5")

e7.5vse9.5deg <- e7.5_e9.5$deg
length(unique(e7.5vse9.5deg$target_id)) #check total number of DE transcripts
length(unique(e7.5vse9.5deg$ens_gene)) #check total number of DE genes

e7.5_e7.5vse9.5 <- e7.5vse9.5deg[e7.5vse9.5deg$log2FC <= -log2(1.5),]
length(unique(e7.5_e7.5vse9.5$target_id)) #check total number of transcripts up-regulated at e7.5, compared to e9.5
length(unique(e7.5_e7.5vse9.5$ens_gene)) #check total number of genes up-regulated at e7.5, compared to e9.5
#write.table(e7.5_e7.5vse9.5, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2A_sleuth/e7.5_e7.5vse9.5_DEtransGenes.txt", row.names = F, quote = F, sep = "\t")

e9.5_e7.5vse9.5 <- e7.5vse9.5deg[e7.5vse9.5deg$log2FC >= log2(1.5),]
length(unique(e9.5_e7.5vse9.5$target_id)) #check total number of transcripts up-regulated at e9.5, compared to e7.5
length(unique(e9.5_e7.5vse9.5$ens_gene)) #check total number of genes up-regulated at e9.5, compared to e7.5
#write.table(e9.5_e7.5vse9.5, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2A_sleuth/e9.5_e7.5vse9.5_DEtransGenes.txt", row.names = F, quote = F, sep = "\t")
```
For **e8.5 vs e9.5** test: <br>
+ Number of DE transcripts: 1255 <br>
+ Number of DE genes: 1124 <br>
+ Number of transcripts up at e8.5: 358 <br>
+ Number of genes up at e8.5: 321 <br>
+ Number of transcripts up at e9.5: 897 <br>
+ Number of genes up at e9.5: 806 <br>
```{r, eval=F}
# e8.5 vs e9.5 test
e8.5_e9.5 <- deg(e8.5vsE9.5, t2g, "e8.5", "e9.5")

e8.5vse9.5deg <- e8.5_e9.5$deg
length(unique(e8.5vse9.5deg$target_id)) #check total number of DE transcripts
length(unique(e8.5vse9.5deg$ens_gene)) #check total number of DE genes

e8.5_e8.5vse9.5 <- e8.5vse9.5deg[e8.5vse9.5deg$log2FC <= -log2(1.5),]
length(unique(e8.5_e8.5vse9.5$target_id)) #check total number of transcripts up-regulated at e8.5, compared to e9.5
length(unique(e8.5_e8.5vse9.5$ens_gene)) #check total number of genes up-regulated at e8.5, compared to e9.5
#write.table(e8.5_e8.5vse9.5, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2A_sleuth/e8.5_e8.5vse9.5_DEtransGenes.txt", row.names = F, quote = F, sep = "\t")

e9.5_e8.5vse9.5 <- e8.5vse9.5deg[e8.5vse9.5deg$log2FC >= log2(1.5),]
length(unique(e9.5_e8.5vse9.5$target_id)) #check total number of transcripts up-regulated at e9.5, compared to e8.5
length(unique(e9.5_e8.5vse9.5$ens_gene)) #check total number of genes up-regulated at e9.5, compared to e8.5
#write.table(e9.5_e8.5vse9.5, "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2A_sleuth/e9.5_e8.5vse9.5_DEtransGenes.txt", row.names = F, quote = F, sep = "\t")
```
This part has function `deg2()` which is for plotting purpose, as for volcano plot we need insignificant transcripts also.
```{r, eval = F}
#function deg2 is for plotting purpose, as for volcano plot I needed insignificant transcripts also
deg2 <-  function(description, targetMap, con1, con2){
  #### 1.2. transcript level
  so1trans <- sleuth_prep(description, target_mapping = targetMap, extra_bootstrap_summary = TRUE)
  so1trans <- sleuth_fit(so1trans, ~condition, 'full')
  so1trans <- sleuth_fit(so1trans, ~1, 'reduced')
  so1trans <- sleuth_lrt(so1trans, 'reduced', 'full')
  
  sleuth_table1trans <- sleuth_results(so1trans, 'reduced:full', 'lrt', show_all = FALSE, pval_aggregate = so1trans$pval_aggregate)
  
  ## finding up/down regulated transcripts
  cond1 <- description$sample[description$condition == con1]
  cond2 <- description$sample[description$condition == con2]
  
  raw_cond1 <- subset(so1trans$obs_raw, so1trans$obs_raw$sample %in% cond1)
  raw_cond2 <- subset(so1trans$obs_raw, so1trans$obs_raw$sample %in% cond2)
  
  #calculate mean TPM of each transcript across all samples
  means_cond1 <- aggregate(tpm~target_id, data = raw_cond1, FUN=function(x) c(mean=mean(x)))
  colnames(means_cond1) <- c("target_id", "TPM_con1")
  
  means_cond2 <- aggregate(tpm~target_id, data = raw_cond2, FUN=function(x) c(mean=mean(x)))
  colnames(means_cond2) <- c("target_id", "TPM_con2")
  
  merged_means <- merge(means_cond1, means_cond2, by = c("target_id"))
  
  sleuth_tableTrans_TPM <- left_join(sleuth_table1trans, merged_means)
  
  #calculate FC between 2 condition
  sleuth_tableTrans_TPM$TPM_con1 <- sleuth_tableTrans_TPM$TPM_con1 + 1
  sleuth_tableTrans_TPM$TPM_con2 <- sleuth_tableTrans_TPM$TPM_con2 + 1
  sleuth_tableTrans_TPM <- transform(sleuth_tableTrans_TPM, log2FC = log2(sleuth_tableTrans_TPM$TPM_con2 / sleuth_tableTrans_TPM$TPM_con1))
  
  return(sleuth_tableTrans_TPM)
}

e7.5_e8.5 <- deg2(e7.5vsE8.5, t2g, "e7.5", "e8.5")
#save(e7.5_e8.5, file = "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2A_sleuth/e7.5_e8.5_allTrans-forPlots.rda")

e7.5_e9.5 <- deg2(e7.5vsE9.5, t2g, "e7.5", "e9.5")
#save(e7.5_e9.5, file = "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2A_sleuth/e7.5_e9.5_allTrans-forPlots.rda")

e8.5_e9.5 <- deg2(e8.5vsE9.5, t2g, "e8.5", "e9.5")
#save(e8.5_e9.5, file = "/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/2A_sleuth/e8.5_e9.5_allTrans-forPlots.rda")
```

## 3. Timepoint-specific genes
Last, we integrate results from hierarchical clustering and DEG analysis to get timepoint-specific genes, defined as the following. <br>
a. E8.5-specific transcripts: transcripts in e8.5 hierarchical cluster, are up-regulated at e8.5 (compared to e7.5) or are up-regulated at e8.5 (compared to e9.5). E8.5-specific genes are ones associated with e8.5-specific transcripts. <br>
b.	E7.5-specific transcripts: transcripts in e7.5 hierarchical cluster, are up-regulated at e7.5 (compared to e9.5), and are not in e8.5-specific group. E7.5-specific genes are ones associated with e7.5-specific transcripts. <br>
c.	E9.5-specific transcripts: transcripts in e9.5 hierarchical cluster, are up-regulated at e9.5 (compared to e7.5), and are not in e8.5-specific group. E9.5-specific genes are ones associated with e9.5-specific transcripts. <br>
Eventually, we should get the number of timepoint-specific transcripts (and genes): <br>
+ e7.5: 1138 transcripts (922 genes) <br>
+ e8.5: 998 transcripts (915 genes) <br>
+ e9.5: 2356 transcripts (1952 genes) <br>
In the following code, we also get TF/co-factor genes specific to each timepoint for downstream analyses. The TF and co-factor was obtained from AnimalTFDB (version 3.0).
```{r}
TF_cofactors <- read.table("Files/mm10_TF_coTF_EnsemblID.txt")

#e8.5 specific genes
e8.5Hier <- read.table("Files/e8.5Group.txt", header = T)
e8.5_e7.5vsE8.5 <- read.table("Files/e8.5_e7.5vsE8.5_DEtransGenes.txt", header = T)
e8.5_e8.5vsE9.5 <- read.table("Files/e8.5_e8.5vse9.5_DEtransGenes.txt", header = T)

e8.5specificTrans <- intersect(union(e8.5_e7.5vsE8.5$target_id, e8.5_e8.5vsE9.5$target_id), unique(e8.5Hier$transcripts))
length(e8.5specificTrans)
#write.table(e8.5specificTrans, "Files/e8.5specific_trans.txt", row.names = F, quote = F)

e8.5specific <- unique(t2g[t2g$target_id %in% e8.5specificTrans, "ens_gene"])
length(e8.5specific)
#write.table(e8.5specific, "Files/e8.5specific_ensGenes.txt", row.names = F, quote = F)

e8.5TFs <- intersect(e8.5specific, TF_cofactors$V1)
length(e8.5TFs)
#write.table(e8.5TFs, "Files/e8.5specific_TFensGenes.txt", row.names = F, quote = F)
```

```{r}
#e7.5 specific genes
e7.5Hier <- read.table("Files/e7.5Group.txt", header = T)
e7.5_e7.5vsE9.5 <- read.table("Files/e7.5_e7.5vse9.5_DEtransGenes.txt", header = T)

e7.5specificTrans <- setdiff(intersect(e7.5_e7.5vsE9.5$target_id, e7.5Hier$transcripts), e8.5specificTrans)
length(e7.5specificTrans)
#write.table(e7.5specificTrans, "Files/e7.5specific_trans.txt", row.names = F, quote = F)

e7.5specific <- unique(t2g[t2g$target_id %in% e7.5specificTrans, "ens_gene"])
length(e7.5specific)
#write.table(e7.5specific, "Files/e7.5specific_ensGenes.txt", row.names = F, quote = F)

e7.5TFs <- intersect(e7.5specific, TF_cofactors$V1)
length(e7.5TFs)
#write.table(e7.5TFs, "Files/e7.5specific_TFensGenes.txt", row.names = F, quote = F)
```
```{r}
#e9.5 specific genes
e9.5Hier <- read.table("Files/e9.5Group.txt", header = T)
e9.5_e7.5vsE9.5 <- read.table("Files/e9.5_e7.5vse9.5_DEtransGenes.txt", header = T)

e9.5specificTrans <- setdiff(intersect(e9.5_e7.5vsE9.5$target_id, e9.5Hier$transcripts), e8.5specificTrans)
length(e9.5specificTrans)
#write.table(e9.5specificTrans, "Files/e9.5specific_trans.txt", row.names = F, quote = F)

e9.5specific <- unique(t2g[t2g$target_id %in% e9.5specificTrans, "ens_gene"])
length(e9.5specific)
#write.table(e9.5specific, "Files/e9.5specific_ensGenes.txt", row.names = F, quote = F)

e9.5TFs <- intersect(e9.5specific, TF_cofactors$V1)
length(e9.5TFs)
#write.table(e9.5TFs, "Files/e9.5specific_TFensGenes.txt", row.names = F, quote = F)
```

Gene ontology analyses for time-point specific genes:
```{r}
library("clusterProfiler")
library("org.Mm.eg.db")

go <- function(subnetworkGenes) {
  GO <- enrichGO(gene = subnetworkGenes, OrgDb=org.Mm.eg.db, ont = "BP", qvalueCutoff = 1, pvalueCutoff = 1, maxGSSize=1000, readable = T, keyType = "ENSEMBL")
  goSimplify <- GO
  simplify2 <- as.data.frame(goSimplify)
  simplify2$GeneRatio <- sapply(strsplit(simplify2$GeneRatio, "/"), function(x) as.numeric(x[1])/as.numeric(x[2]))
  simplify2$BgRatio <- sapply(strsplit(simplify2$BgRatio, "/"), function(x) as.numeric(x[1])/as.numeric(x[2]))
  simplify2$Fold <- as.numeric(simplify2$GeneRatio)/as.numeric(simplify2$BgRatio)
  simplify2 <- simplify2[simplify2$qvalue <= 0.05 & simplify2$Fold >= 2 & simplify2$Count >= 5,]
  
  return(simplify2)
}

#specific genes
for (i in c("e7.5", "e8.5", "e9.5")) {
  genes <- read.table(paste0("/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/PlacentaRNA-seq/Files/", i, "specific_TSS.bed"), header = F)
  res <- go(genes$V4)
  write.table(res, paste0("/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/PlacentaRNA-seq/Files/additionalAnalyses/", i, "specific_BP_filtered.txt"), quote = F, sep = "\t", row.names = F)
}
```
E7.5-specific genes in general, and e7.5-specific genes that are unique to e7.5 timepoint (not present in any other timepoint).
```{r}
e8.5 <- read.table("Files/e8.5specific_ensGenes.txt", header = F)
e7.5 <- read.table("Files/e7.5specific_ensGenes.txt", header = F)
e9.5 <- read.table("Files/e9.5specific_ensGenes.txt", header = F)

trueE7.5 <- setdiff(setdiff(e7.5$V1, e8.5$V1), e9.5$V1)
length(e7.5$V1)
length(trueE7.5)
trueE7.5go <- go(trueE7.5)
write.table(trueE7.5go, "Files/trueE7.5go.txt", sep = "\t", quote = F, row.names = F)
multiE7.5 <- setdiff(e7.5$V1, trueE7.5)
intersect(multiE7.5, e8.5$V1)
intersect(multiE7.5, e9.5$V1)


e7.5specificTrans <- read.table("Files/e7.5specific_trans.txt", header = F)
#write.table(t2g[t2g$ens_gene %in% multiE7.5 & t2g$target_id %in% e7.5specificTrans$V1,], "Files/multiE7.5.txt", sep = "\t", quote = F, row.names = F)
multiE7.5go <- go(multiE7.5)
#write.table(multiE7.5go, "Files/multiE7.5go.txt", sep = "\t", quote = F, row.names = F)

i<-"e7.5"
genes <- read.table(paste0("/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/PlacentaRNA-seq/Files/", i, "specific_TSS.bed"), header = F)
res <- go(genes$V4)
setdiff(trueE7.5go$Description, res$Description)
setdiff(res$Description, trueE7.5go$Description)
```
E8.5-specific genes in general, and e8.5-specific genes that are unique to e8.5 timepoint (not present in any other timepoint).
```{r}
trueE8.5 <- setdiff(setdiff(e8.5$V1, e7.5$V1), e9.5$V1)
length(e8.5$V1)
length(trueE8.5)
trueE8.5go <- go(trueE8.5)
write.table(trueE8.5go, "Files/trueE8.5go.txt", sep = "\t", quote = F, row.names = F)
multiE8.5 <- setdiff(e8.5$V1, trueE8.5)
intersect(multiE8.5, e7.5$V1)
intersect(multiE8.5, e9.5$V1)

e8.5specificTrans <- read.table("Files/e8.5specific_trans.txt", header = F)
write.table(t2g[t2g$ens_gene %in% multiE8.5 & t2g$target_id %in% e8.5specificTrans$V1,], "Files/multiE8.5.txt", sep = "\t", quote = F, row.names = F)
multiE8.5go <- go(multiE8.5)
#write.table(multiE8.5go, "Files/multiE8.5go.txt", sep = "\t", quote = F, row.names = F)

i<-"e8.5"
genes <- read.table(paste0("/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/PlacentaRNA-seq/Files/", i, "specific_TSS.bed"), header = F)
res <- go(genes$V4)
setdiff(trueE8.5go$Description, res$Description)
setdiff(res$Description, trueE8.5go$Description) #only here has labyrinth terms

length(intersect(e8.5$V1, e7.5$V1))
length(intersect(e8.5$V1, e9.5$V1))
```
E9.5-specific genes in general, and e9.5-specific genes that are unique to e9.5 timepoint (not present in any other timepoint).
```{r}
trueE9.5 <- setdiff(setdiff(e9.5$V1, e8.5$V1), e7.5$V1)
length(e9.5$V1)
length(trueE9.5)
trueE9.5go <- go(trueE9.5)
write.table(trueE9.5go, "Files/trueE9.5go.txt", sep = "\t", quote = F, row.names = F)
multiE9.5 <- setdiff(e9.5$V1, trueE9.5)
intersect(multiE9.5, e7.5$V1)
intersect(multiE9.5, e8.5$V1)

e9.5specificTrans <- read.table("Files/e9.5specific_trans.txt", header = F)
write.table(t2g[t2g$ens_gene %in% multiE9.5 & t2g$target_id %in% e9.5specificTrans$V1,], "Files/multiE9.5.txt", sep = "\t", quote = F, row.names = F)
#write.table(distinct(t2g[t2g$ens_gene %in% multiE9.5, 2:3]), "Files/multiE9.5.txt", sep = "\t", quote = F, row.names = F)

multiE9.5go <- go(multiE9.5)
write.table(multiE9.5go, "Files/multiE9.5go.txt", sep = "\t", quote = F, row.names = F)

multiE9.5_e8.5go <- go(intersect(multiE9.5, e8.5$V1))
write.table(multiE9.5_e8.5go, "Files/multiE9.5_e8.5go.txt", sep = "\t", quote = F, row.names = F)

i<-"e9.5"
genes <- read.table(paste0("/work/LAS/geetu-lab/hhvu/project1_2/rna-seq/PlacentaRNA-seq/Files/", i, "specific_TSS.bed"), header = F)
res <- go(genes$V4)
setdiff(trueE9.5go$Description, res$Description)
setdiff(res$Description, trueE9.5go$Description)
```

```{r}
sessionInfo()
```