07-NONTARGETS_06.Rmd

# GO and KEGG: non targets 06

## Load the data

```{r}
# load table with non_targets (binding probability > 0.6)
data_MAPP <- read.table("data/inclusion_tables/0.6/non_targets.tsv", 
                        header = T)

# select control and cancer columns
CTRL <- data_MAPP %>% dplyr::select_if(., grepl("^CONTROL_", names(.))) 
CANCER <- data_MAPP %>% dplyr::select_if(., grepl("^CANCER_", names(.))) 

# use row_t_welch instead of usual t.test function since t.test produces 
# errors in some cases: t.test error: data are essentially constant
# NAs are assigned for problematic cases 
# https://stats.stackexchange.com/questions/499634/t-test-error-data-are-essentially-constant/499637
# library matrixTests
data_MAPP <- data_MAPP %>%
  mutate(padj = p.adjust(row_t_welch(CTRL, CANCER)$pvalue,
                         method = "BH", n=nrow(data_MAPP)))
```

```{r}
# view table interactively
DT::datatable(data_MAPP, options = list(scrollX = TRUE))
```

## Gene ontology (GO)

### Prepare the gene list 

```{r}
# select genes from data_MAPP table based on padj values
# threshold is set in index.Rmd: genes_padj_cutoff
genes <- na.omit(data_MAPP) %>% arrange(desc(ABSDIFF)) %>% 
  filter(padj < genes_padj_cutoff) %>% dplyr::select(gene_id) %>% unlist() 
genes <- unlist(strsplit(genes, ","))
```

```{r}
# preview the gene names
head(genes)
```

### Biological process (BP)

```{r}
# select type of GO analysis
type_of_GO = "BP"

# create an object with a name based on the type of analysis
assign(paste0("GO_", type_of_GO), enrichGO(
  genes,
  organismDB,
  keyType = "ENSEMBL",
  ont = type_of_GO,
  pvalueCutoff = pvalueCutoff,
  minGSSize = minGSSize,
  pAdjustMethod = "BH",
  universe = universe,
  readable = TRUE))

# assign the object created above to "df_GO" object
df_GO <- eval(parse(text = paste0("GO_", type_of_GO)))
```

```{r}
# if there are enriched terms in "df_GO" object, create a table with the results
# filtered by specified p-adjusted threshold, if no enriched terms were found
# print "no GOs were found"
if(nrow(df_GO) > 0){
  DT::datatable(df_GO@result %>% filter(p.adjust < pvalueCutoff), 
                options = list(scrollX = TRUE))
} else "no GOs were found"
```

```{r}
# if there are enriched terms in "df_GO" object, create a barplot with 
# specified number of GOs to plot, if no enriched terms were found
# print "no GOs to plot"
if(nrow(df_GO) > 0){
  barplot(df_GO, showCategory = showCategory, 
                                title = type_of_GO)
} else "no GOs to plot"
```

### Molecular function (MF)

```{r}
# select type of GO analysis
type_of_GO = "MF"

# create an object with a name based on the type of analysis
assign(paste0("GO_", type_of_GO), enrichGO(
  genes,
  organismDB,
  keyType = "ENSEMBL",
  ont = type_of_GO,
  pvalueCutoff = pvalueCutoff,
  minGSSize = minGSSize,
  pAdjustMethod = "BH",
  universe = universe,
  readable = TRUE))

# assign the object created above to "df_GO" object
df_GO <- eval(parse(text = paste0("GO_", type_of_GO)))
```

```{r}
# if there are enriched terms in "df_GO" object, create a table with the results
# filtered by specified p-adjusted threshold, if no enriched terms were found
# print "no GOs were found"
if(nrow(df_GO) > 0){
  DT::datatable(df_GO@result %>% filter(p.adjust < pvalueCutoff), 
                options = list(scrollX = TRUE))
} else "no GOs were found"
```

```{r}
# if there are enriched terms in "df_GO" object, create a barplot with 
# specified number of GOs to plot, if no enriched terms were found
# print "no GOs to plot"
if(nrow(df_GO) > 0){
  barplot(df_GO, showCategory = showCategory, 
                                title = type_of_GO)
} else "no GOs to plot"
```

### Cellular component (CC)

```{r}
# select type of GO analysis
type_of_GO = "CC"

# create an object with a name based on the type of analysis
assign(paste0("GO_", type_of_GO), enrichGO(
  genes,
  organismDB,
  keyType = "ENSEMBL",
  ont = type_of_GO,
  pvalueCutoff = pvalueCutoff,
  minGSSize = minGSSize,
  pAdjustMethod = "BH",
  universe = universe,
  readable = TRUE))

# assign the object created above to "df_GO" object
df_GO <- eval(parse(text = paste0("GO_", type_of_GO)))
```

```{r}
# if there are enriched terms in "df_GO" object, create a table with the results
# filtered by specified p-adjusted threshold, if no enriched terms were found
# print "no GOs were found"
if(nrow(df_GO) > 0){
  DT::datatable(df_GO@result %>% filter(p.adjust < pvalueCutoff), 
                options = list(scrollX = TRUE))
} else "no GOs were found"
```

```{r}
# if there are enriched terms in "df_GO" object, create a barplot with 
# specified number of GOs to plot, if no enriched terms were found
# print "no GOs to plot"
if(nrow(df_GO) > 0){
  barplot(df_GO, showCategory = showCategory, 
                                title = type_of_GO)
} else "no GOs to plot"
```

## KEGG

### Prepare the the data

```{r, warning = FALSE}
# select gene_id in ENSEMBL format and DIFF from the result table
DIFF <- data_MAPP %>% dplyr::select(gene_id, DIFF)

# rename the gene_id column  
names(DIFF)[1] <- "ENSEMBL"

# preview the head of the table
head(DIFF)

# create new df by converting ENSEMBL ID into ENTREZID and adding DIFF column
# arrange table in descending order by abs(DIFF)
genes_ENTREZ_df <- 
  bitr(na.omit(data_MAPP$gene_id[data_MAPP$padj < genes_padj_cutoff]), 
       fromType = "ENSEMBL", toType = "ENTREZID", OrgDb= organismDB) %>% 
  left_join(DIFF, by = "ENSEMBL") %>% arrange(desc(abs(DIFF)))

# preview obtained table
head(genes_ENTREZ_df )

# create a vector with DIFF values
genes_DIFF <- genes_ENTREZ_df$DIFF
genes_DIFF

# add names based on ENTREZID
names(genes_DIFF) <- genes_ENTREZ_df$ENTREZID

# preview genes in ENTREZ format with corresponding fold changes
head(genes_DIFF)
```

### run enrichKEGG

```{r}
# run enrichKEGG using options specified in "index.Rmd" file
KEGGresults <- enrichKEGG(
  names(genes_DIFF),
  organism = "hsa",
  keyType = "ncbi-geneid",
  pvalueCutoff = pvalueCutoff,
  pAdjustMethod = "BH",
  universe = as.character(na.omit(ENTREZ_universe)),
  use_internal_data = FALSE
)
```

```{r}
# if there are enriched KEGG pathways found in "KEGGresults" object, 
# create a table with the results filtered by specified p-adjusted threshold, 
# if no enriched terms were found, print "no KEGG pathways were found"
if (!is.null(KEGGresults)) {
if(nrow(KEGGresults@result %>% filter(p.adjust < pvalueCutoff)) > 0){
  DT::datatable(KEGGresults@result %>% filter(p.adjust < pvalueCutoff), 
              options = list(scrollX = TRUE))}
} else "no KEGG pathways were found"
```