25_network_analysis.Rmd

---
title: "Metabolic network analysis"
author: "Sergio Picart-Armada"
date: "8th March, 2020"
output:
  html_document:
    toc: TRUE
    toc_float: true
    code_folding: hide
    df_print: paged
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, message = FALSE, 
                      error = FALSE, warning = FALSE, 
                      fig.height = 4)
```

# Loading the data

```{r cars}
library(plyr)
library(dplyr)
library(tidyr)
library(magrittr)

library(ggplot2)
library(ggpmisc)
library(plotly)
library(psych)

library(data.table)

library(clusterProfiler)

# metabolomics
library(FELLA)
library(igraph)

config <- new.env()
source("config.R", local = config)
source("helpers.R")

theme_set(theme_bw())

dir.out <- config$dir.fella
if (!dir.exists(dir.out)) dir.create(dir.out)

set.seed(1)
```

FELLA database

```{r}
# save db in project root
fella.db <- config$fella.db

if (!dir.exists(fella.db)) {
  g <- FELLA::buildGraphFromKEGGREST(organism = "mmu", filter.path = "01100")
  FELLA::buildDataFromGraph(
    g, databaseDir = fella.db,
    internalDir = FALSE,
    matrices = c("hypergeom", "diffusion"),
    normality = c("hypergeom", "diffusion"),
    niter = 1000)
}
fella.data <- FELLA::loadKEGGdata(fella.db, internalDir = FALSE, loadMatrix = "hypergeom")
```

```{r}
fc.ref <- config$fc.ref
q.ref <- config$q.ref
```

Load fold changes

```{r}
df.fc.trt <- read.csv(config$df.fc.trt)
```

# FELLA network analysis

## Network summary

Mouse network from KEGG

```{r}
fella.data
```
## Lists of metabolites

```{r}
cpd.young <- filter(df.fc.trt, Age == "Young" & se == "Metabolites" & signif)$db.id %>% unique %>% na.omit
cpd.old <- filter(df.fc.trt, Age == "Old" & se == "Metabolites" & signif)$db.id %>% unique %>% na.omit
cpd.both <- intersect(cpd.old, cpd.young)


cpd.bkgd <- filter(df.fc.trt, Age == "Old" & se == "Metabolites")$db.id %>% unique %>% na.omit
```


## Young

```{r}
enr.young <- FELLA::enrich(cpd.young, methods = "diffusion", data = fella.data)
g.young <- FELLA::generateResultsGraph(object = enr.young, data = fella.data)

g.young %>%
  igraph_to_visnetwork()
```

Number of nodes

```{r}
vcount(g.young)
```

Entities in it

```{r}
tab.young <- generateResultsTable(object = enr.young, data = fella.data) 

write.csv(tab.young, paste0(dir.out, "/subnetwork_all_young.csv"))

DT::datatable(tab.young)
```

Enzymes with gene annotations

```{r}
enz.young <- generateEnzymesTable(object = enr.young, data = fella.data) 

write.csv(enz.young, paste0(dir.out, "/subnetwork_enzymes_young.csv"))

DT::datatable(enz.young)
```


## Old


```{r}

enr.old <- FELLA::enrich(cpd.old, methods = "diffusion", data = fella.data)
g.old <- FELLA::generateResultsGraph(object = enr.old, data = fella.data)

g.old %>%
  igraph_to_visnetwork()
```


Number of nodes

```{r}
vcount(g.old)
```

Entities in it

```{r}
tab.old <- generateResultsTable(object = enr.old, data = fella.data) 

write.csv(tab.old, paste0(dir.out, "/subnetwork_all_old.csv"))

DT::datatable(tab.old)
```

Enzymes with gene annotations

```{r}
enz.old <- generateEnzymesTable(object = enr.old, data = fella.data) 

write.csv(enz.old, paste0(dir.out, "/subnetwork_enzymes_old.csv"))

DT::datatable(enz.old)
```

## Intersection old and young

```{r, fig.width=7, fig.height=7}
enr.both <- FELLA::enrich(cpd.both, methods = "diffusion", data = fella.data)
g.both <- FELLA::generateResultsGraph(object = enr.both, data = fella.data, LabelLengthAtPlot = 100)

g.both %>% igraph_to_visnetwork() 
  
```

For exporting: really large plot (first right click and save PNG, then browse it)

```{r}
g.both %>% igraph_to_visnetwork(width = "3200px", height = "3200px")
```


Number of nodes

```{r}
vcount(g.both)
```

Entities in it

```{r}
tab.both <- generateResultsTable(object = enr.both, data = fella.data, LabelLengthAtPlot = 100) 

write.csv(tab.both, paste0(dir.out, "/subnetwork_all_intersectoldyoung.csv"))

DT::datatable(tab.both)
```

Enzymes with gene annotations

```{r}
enz.both <- generateEnzymesTable(object = enr.both, data = fella.data) 

write.csv(enz.both, paste0(dir.out, "/subnetwork_enzymes_intersectoldyoung.csv"))

DT::datatable(enz.both)
```


### Export plot with customised igraph plotting

Can we export to SVG?

```{r, fig.width=8, fig.height=8}
# reproducible layout?
set.seed(2)
g.both.plt <- igraph_add_graphical_params(g.both) 

V(g.both.plt)[com %in% c(4)]$label <- ""
V(g.both.plt)$label.dist <- .7
V(g.both.plt)$label.cex <- .7
# V(g.both.plt)$shape <- "rectangle"
V(g.both.plt)$label.color %<>% paste0("FF")
V(g.both.plt)$color %<>% paste0("55")

pdf(paste0(dir.out, "/subnetwork_intersect.pdf"), width = 10, height = 10)
plot(g.both.plt)
dev.off()
```

### ggnet2

```{r}
# library(network)
# library(ggnet2)
# not in easybuild/cran
```

### NetworkD3

Not easy to export...

```{r}
d3.both.plt <- networkD3::igraph_to_networkD3(g.both.plt, group = V(g.both.plt)$com) 

# Create force directed network plot
networkD3::forceNetwork(
  Links = d3.both.plt$links, Nodes = d3.both.plt$nodes,
  Source = 'source', Target = 'target', NodeID = 'name',
  Group = 'group')

```



# Reproducibility

```{r}
date()
```

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