PTM_analysis.Rmd

---
title: "Post translation modification"
author: "Briana Mittleman"
date: "3/6/2020"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

```{r}
library(ggpubr)
library(workflowr)
library(tidyverse)
```

Looking at protien interactions. https://downloads.thebiogrid.org/BioGRID/Release-Archive/BIOGRID-3.5.182/

Are genes with dAPA more likely to be in genes with known protein protein interactions. Given the regulatory elements in the 3' UTR. This could help us understand a mechanism. 

( interactions, chemical associations, and post-translational modifications (PTM))

```{bash,eval=F}
mkdir ../data/bioGRID
```



Look at data:  

Fix colnames  
```{r}
Biogrid=read_tsv("../data/bioGRID/BIOGRID-ORGANISM-Homo_sapiens-3.5.182.tab2.txt")

colnames(Biogrid)= c( "BioGRID_Interaction_ID",  "Entrez_Gene_Interactor_A", "Entrez_Gene_Interactor_B", "BioGRID_ID_Interactor_A",  "BioGRID_ID_Interactor_B"   ,   "Systematic_Name_Interactor_A","Systematic_Name_Interactor_B", "Official_Symbol_Interactor_A", "Official_Symbol_Interactor_B","Synonyms_Interactor_A", "Synonyms_Interactor_B" , "Experimental_System", "Experimental_System_Type" ,"Author" , "Pubmed_ID" ,"Organism_Interactor_A", "Organism_Interactor_B", "Throughput","Score", "Modification" , "Phenotypes","Qualifications", "Tags" , "Source Database" )


```

Select the official names for the interactors:  

```{r}
Biogridsmall=Biogrid %>% dplyr::select(Official_Symbol_Interactor_A, Official_Symbol_Interactor_B,Score, Modification, Phenotypes, Tags) 
```

I need a way to remove duplicates. I can do this by making unordered sets of these. I will need the uniq sets.

Make a set with the pasted version of A:B and B:A, keep the unique set  

```{r}
BioGridsets=Biogridsmall %>% mutate(Afirst=paste(Official_Symbol_Interactor_A, Official_Symbol_Interactor_B, sep="_:_"), Bfirst=paste(Official_Symbol_Interactor_B, Official_Symbol_Interactor_A, sep="_:_")) 

Allsets= as.data.frame(c(BioGridsets$Afirst, BioGridsets$Bfirst)) %>%  unique()

colnames(Allsets)=c("Interaction")


AllGenes=as.data.frame(c(Biogridsmall$Official_Symbol_Interactor_A, Biogridsmall$Official_Symbol_Interactor_B)) %>% unique()

colnames(AllGenes)=c("Genes")
```
I want to know if my genes are in either of these sets. I also need the set of all genes that are involved. 



```{r}
Allsets_sep= Allsets %>% separate(Interaction, into=c("a","b"), sep="_:_")
```

Get all of the genes together in one column (not unique) , group by the gene and count how many interactions


```{r}
GenesWint= as.data.frame(c(Allsets_sep$a, Allsets_sep$b))
colnames(GenesWint)= c("gene")
GenesWint_g= GenesWint %>% group_by(gene) %>% summarise(nInt=n())


```


Join this with the genes I test. 

```{r}
NucRes=read.table("../data/DiffIso_Nuclear_DF/AllPAS_withGeneSig.txt", header = T, stringsAsFactors = F) %>% group_by(gene, SigPAU2) %>% summarise(N=n()) %>% spread(SigPAU2,N) %>% replace_na(list(Yes=0)) %>% mutate(dAPA=ifelse(Yes>=1, "Yes", "No")) %>% dplyr::select(-Yes, -No)

NucResAll= NucRes  %>% left_join(GenesWint_g, by="gene") %>%  replace_na(list(nInt=0)) 

write.table(GenesWint_g, "../data/bioGRID/GeneswInteractions.txt",col.names = T, row.names = F, quote = F)
```


```{r}
ggplot(NucResAll,aes(x=dAPA, y=log10(nInt +1))) + geom_boxplot() + stat_compare_means()
```

Does not look like this can explain the differences.  

Enriched for non 0?



```{r}
NucResAll_g= NucResAll %>% mutate(HasInteraction=ifelse(nInt>0, "Yes", "No")) %>% group_by(dAPA, HasInteraction) %>% summarise(nWithSet=n())

NucResAll_g
```


```{r}
5581/(5581+814)
1869/(1869+280)
```

Doesnt look like differentail are more likely to have an interaction.  


##filter for modifications  

```{r}
Biogridsmall %>% dplyr::select(Modification) %>% unique()
```