03_foldseek_distances_network.Rmd

---
title: "03_domainome_structural_network"
author: "Toni Beltran"
date: "28/02/2024"
output: html_document
---


```{r load data}

library(data.table)

base_dir="/path/to/your/files"
setwd(base_dir)

foldseek_distances<-fread("analysis_files/pdb_files/foldseek_easy_allvsall")
foldseek_distances[,distance:=1-V6]

colnames(foldseek_distances)<-c("Source","Target","V3","V4","V5","Weight","Distance")
foldseek_distances$Type<-rep("Undirected",times=nrow(foldseek_distances))

nodes<-data.table(Id=unique(foldseek_distances$Source))
nodes$PFAM_ID<-unlist(lapply(nodes$Id,FUN=function(string){
  return(strsplit(string,"_")[[1]][2])
}))


#paint by SCOP class
pfam_to_scop_class<-fread("analysis_files/PFAM_ID_to_SCOP_class.tsv")
nodes<-merge(nodes,pfam_to_scop_class,by="PFAM_ID")


#write node and edge lists to load into gephi
write.table(foldseek_distances[Source!=Target,c("Source","Target","Type","Weight")],
            file="output_files/fseek_distances_edgelist.csv",sep=",",row.names = FALSE)

write.table(nodes[,c("Id","scop_class")],file="output_files/fseek_distances_nodelist.csv",sep=",",row.names = FALSE)



```