swarmPlots_QTLs.Rmd

---
title: "Top APA QTL in other phenotypes"
author: "Briana Mittleman"
date: "10/8/2018"
output: html_document
---

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

##Upload Data:  

Library
```{r}
library(workflowr)
library(reshape2)
library(tidyverse)
library(VennDiagram)
library(data.table)
library(cowplot)
```


Permuted Results from APA:

```{r}
nuclearAPA=read.table("../data/perm_QTL_trans/filtered_APApeaks_merged_allchrom_refseqGenes_pheno_Nuclear_transcript_permResBH.txt", stringsAsFactors = F, header = T)
totalAPA=read.table("../data/perm_QTL_trans/filtered_APApeaks_merged_allchrom_refseqGenes_pheno_Total_transcript_permResBH.txt", stringsAsFactors = F, header=T)  
```



I want to use a buzz swarm plot to plot peak usage for some of the top QTLs. I can use the examples I gave Tony. 


Nuclear:  
* peak305794, sid: 7:128635754  

* peak: 164036, sid: 2:3502035

Total:  

* **Peak: peak228606, SID 3:150302010**

* Peak: peak152751, SID 19:4236475


I need to pull out the genotypes for each snp and the corresponding phenotype. I want to make a python script that I can give a snp and a peak and it will make a table with the genotypes and phenotypes for the necessary gene snp pair.  


##Example Peak: peak228606, SID 3:150302010  
```{r}
geno3_m=fread("../data/apaExamp/geno3_150302010.txt", header = T) %>% dplyr::select(starts_with("NA")) %>% t
geno3df= data.frame(geno3_m) %>% separate(geno3_m, into=c("geno", "dose", "extra"), sep=":") %>% dplyr::select(dose) %>% rownames_to_column(var="ind")
apaphen228606_m= fread("../data/apaExamp/Total.peak228606.txt", header = T) %>% dplyr::select(starts_with("NA")) %>% t
apaphen228606_df=data.frame(apaphen228606_m) %>% rownames_to_column(var="ind")

```

```{r}
toplotAPA=geno3df %>% inner_join(apaphen228606_df, by="ind")
toplotAPA$dose= as.factor(toplotAPA$dose)
colnames(toplotAPA)= c("ind", "Genotype", "APA")
EIF2A_APAex=ggplot(toplotAPA, aes(y=APA, x=Genotype, by=Genotype, fill=Genotype)) + geom_boxplot() + geom_jitter() + labs(y="APA phenotype", title="Total APA: Peak 228606, EIF2A") + scale_fill_brewer(palette="YlOrRd")
ggsave("../output/plots/EIF2a_APA.png", EIF2A_APAex)
```

This is in the gene EIF2A, I need to find this in the eQTL data. The ensg id for this gene is ENSG00000144895.  

```{r}
RNAseqEIF2A_m=read.table("../data/apaExamp/RNAseq.phenoEIF2A.txt", header=T) %>% dplyr::select(starts_with("NA")) %>% t 
RNAseqEIF2A_df= data.frame(RNAseqEIF2A_m) %>% rownames_to_column("ind")

plotRNA=geno3df %>% inner_join(RNAseqEIF2A_df, by="ind")
plotRNA$dose= as.factor(plotRNA$dose)
colnames(plotRNA)= c("ind", "Genotype", "Expression")

EIF2A_RNAex=ggplot(plotRNA, aes(y=Expression, x=Genotype, by=Genotype, fill=Genotype)) + geom_boxplot() + geom_jitter() + labs(y="Normalized Expression", title="Gene Expression: EIF2A") + scale_fill_brewer(palette="YlGn")

ggsave("../output/plots/EIF2a_RNA.png", EIF2A_RNAex)

```

Try this in protein:  

```{r}
ProtEIF2A_m=read.table("../data/apaExamp/ProtEIF2A.txt", header=T) %>% dplyr::select(starts_with("NA")) %>% t 
ProtEIF2A_df= data.frame(ProtEIF2A_m) %>% rownames_to_column("ind")

plotProt=geno3df %>% inner_join(ProtEIF2A_df, by="ind")
plotProt$dose= as.factor(plotProt$dose)
colnames(plotProt)= c("ind", "Genotype", "Prot_level")

IF2A_Protex= ggplot(plotProt, aes(y=Prot_level, x=Genotype, by=Genotype, fill=Genotype)) + geom_boxplot() + geom_jitter() + labs(y="Normalized Protein Level", title="Protein Level: EIF2A") +scale_fill_brewer(palette="PuBu")

ggsave("../output/plots/EIF2a_Prot.png", IF2A_Protex)

```


```{r}
multphenoEIF2a=plot_grid(EIF2A_APAex,IF2A_Protex,EIF2A_RNAex,nrow=1)
ggsave("../output/plots/EIF2a_multpheno.png", multphenoEIF2a, width=15, height=5)
```



Do this with 4su 60:  

have to remove the #

```{r}
su60_EIF2A_m=read.table("../data/apaExamp/Foursu60EIF2A.txt", header=T) %>% dplyr::select(starts_with("NA")) %>% t 
su60_EIF2A_df= data.frame(su60_EIF2A_m) %>% rownames_to_column("ind")

plot4su60=geno3df %>% inner_join(su60_EIF2A_df, by="ind")
plot4su60$dose= as.factor(plot4su60$dose)
colnames(plot4su60)= c("ind", "Genotype", "su60")

EIF2A_4su60ex=ggplot(plot4su60, aes(y=su60, x=Genotype, by=Genotype, fill=Genotype)) + geom_boxplot() + geom_jitter() + labs(y="4su60", title="4su 60min Value: EIF2A") + scale_fill_brewer(palette="RdPu") +  theme_classic()

ggsave("../output/plots/EIF2a_4su60.png", EIF2A_4su60ex)

```


Geuvadis RNA  

```{r}
rnaG_EIF2A_m=read.table("../data/apaExamp/RNA_GEU_EIF2A.txt", header=T) %>% dplyr::select(starts_with("NA")) %>% t 
rnaG_EIF2A_df= data.frame(rnaG_EIF2A_m) %>% rownames_to_column("ind")

plotRNAg=geno3df %>% inner_join(rnaG_EIF2A_df, by="ind")
plotRNAg$dose= as.factor(plotRNAg$dose)
colnames(plotRNAg)= c("ind", "Genotype", "RNAg")

EIF2A_RNAgex=ggplot(plotRNAg, aes(y=RNAg, x=Genotype, by=Genotype, fill=Genotype)) + geom_boxplot() + geom_jitter() + labs(y="RNA expression", title="RNA Expression Geuvadis: EIF2A") + scale_fill_brewer(palette="RdPu")

ggsave("../output/plots/EIF2a_RNAg.png", EIF2A_RNAgex)

```

Ribo: 

```{r}
ribo_EIF2A_m=read.table("../data/apaExamp/Ribo_EIF2A.txt", header=T) %>% dplyr::select(starts_with("NA")) %>% t 
ribo_EIF2A_df= data.frame(ribo_EIF2A_m) %>% rownames_to_column("ind")

plotrib=geno3df %>% inner_join(ribo_EIF2A_df, by="ind")
plotrib$dose= as.factor(plotrib$dose)
colnames(plotrib)= c("ind", "Genotype", "Ribo")

EIF2A_riboex=ggplot(plotrib, aes(y=Ribo, x=Genotype, by=Genotype, fill=Genotype)) + geom_boxplot() + geom_jitter() + labs(y="RNA expression", title="Ribo Geuvadis: EIF2A") + scale_fill_brewer(palette="RdPu")

ggsave("../output/plots/EIF2a_Ribo.png", EIF2A_riboex)

```


##Create a script to make the relevent files  


Python script that take a chromosome, snp, peak#, fraction

createQTLsnpAPAPhenTable.py
```{bash,eval=F}
def main(PhenFile, GenFile, outFile, snp, peak):
    fout=open(outFile, "w")
    Phen=open(PhenFile, "r")
    Gen=open(GenFile, "r")
    #get ind and pheno info
    for num, ln in enumerate(Phen):
        if num == 0:
            indiv= ln.split()[4:]
        else:
            id=ln.split()[3].split(":")[3]
            peakID=id.split("_")[2]
            if peakID == peak:
                pheno_list=ln.split()[4:]
                pheno_data=list(zip(indiv,pheno_list))
    pheno_df=pd.DataFrame(data=pheno_data,columns=["Ind", "Pheno"])
    
    for num, lnG in enumerate(Gen):
        if num == 13:
            Ind_geno=lnG.split()[9:]
        if num >= 14: 
            sid= lnG.split()[2]
            if sid == snp: 
                gen_list=lnG.split()[9:]
                allele1=[]
                allele2=[]
                for i in gen_list:
                    genotype=i.split(":")[0]
                    allele1.append(genotype.split("|")[0])
                    allele2.append(genotype.split("|")[1])
          #now i have my indiv., phen, allele 1, alle 2     
                geno_data=list(zip(Ind_geno, allele1, allele2))
    geno_df=pd.DataFrame(data=geno_data, columns=["Ind", "Allele1", "Allele2"])
    full_df=pd.merge(geno_df, pheno_df, how="inner", on="Ind")
    full_df.to_csv(fout, sep="\t", encoding='utf-8', index=False)
    fout.close()
    

if __name__ == "__main__":
    import sys
    import pandas as pd
    chrom=sys.argv[1]
    snp = sys.argv[2]
    peak = sys.argv[3]
    fraction=sys.argv[4]
    
    PhenFile = "/project2/gilad/briana/threeprimeseq/data/phenotypes_filtPeakTranscript/filtered_APApeaks_merged_allchrom_refseqGenes.Transcript_sm_quant.%s.pheno_fixed.txt.gz.phen_chr%s"%(fraction, chrom)
    GenFile= "/project2/gilad/briana/YRI_geno_hg19/chr%s.dose.filt.vcf"%(chrom)
    outFile = "/project2/gilad/briana/threeprimeseq/data/ApaQTL_examples/qtlSNP_PeakAPA%s.%s%s.txt"%(fraction, snp, peak)
    main(PhenFile, GenFile, outFile, snp, peak)
    
```

Use the results to plot the nuclear pheno:  

```{r}
EIF2a_APAnuc=read.table("../data/apaExamp/qtlSNP_PeakAPANuclear.3:150302010peak228606.txt", header=T, stringsAsFactors = F) %>% mutate(Geno=Allele1 + Allele2)

EIF2a_APAnuc$Geno= as.factor(as.character(EIF2a_APAnuc$Geno))


ggplot(EIF2a_APAnuc, aes(y=Pheno, x=Geno, by=Geno, fill=Geno)) + geom_boxplot() + geom_jitter() + labs(y="APA Nuc Usage", title="APA nuc: EIF2A") + scale_fill_brewer(palette="RdPu")
```



This does the total and nuclear fraction of APA. I will do this for a snp and gene and get all of the other phenotypes. This will be similar other than changing the names of the genes and seperating the name for all but protein. 

createQTLsnpMolPhenTable.py  
```{bash,eval=F}
def main(PhenFile, GenFile, outFile, snp, gene, molPhen):
    genenames=open("/project2/gilad/briana/genome_anotation_data/ensemble_to_genename.txt", "r" )
    for ln in genenames:
        geneName=ln.split()[1]
        if geneName == gene:
            gene_ensg=ln.split()[0]
    fout=open(outFile, "w")
    Phen=open(PhenFile, "r")
    Gen=open(GenFile, "r")
    #get ind and pheno info
    for num,ln in enumerate(Phen):
        if num == 0:
            indiv= ln.split()[4:]
        else:
            if molPhen=="Prot":
                gene=ln.split()[3]
                if gene == gene_ensg:
                    pheno_list=ln.split()[4:]
                    pheno_data= list(zip(indiv, pheno_list))
            else:
                full_gene=ln.split()[3]
                gene= full_gene.split(".")[0]
                if gene == gene_ensg:
                    pheno_list=ln.split()[4:]
                    pheno_data= list(zip(indiv, pheno_list))
                    pheno_df=pd.DataFrame(data=pheno_data,columns=["Ind", "Pheno"])
    for num, lnG in enumerate(Gen):
        if num == 13:
            Ind_geno=lnG.split()[9:]
        if num >= 14: 
            sid= lnG.split()[2]
            if sid == snp: 
                gen_list=lnG.split()[9:]
                allele1=[]
                allele2=[]
                for i in gen_list:
                    genotype=i.split(":")[0]
                    allele1.append(genotype.split("|")[0])
                    allele2.append(genotype.split("|")[1])
          #now i have my indiv., phen, allele 1, alle 2     
               geno_data=list(zip(Ind_geno, allele1, allele2))
               geno_df=pd.DataFrame(data=geno_data, columns=["Ind", "Allele1", "Allele2"])
               full_df=pd.merge(geno_df, pheno_df, how="inner", on="Ind")
               full_df.to_csv(fout, sep="\t", encoding='utf-8', index=False)
    fout.close()
    

if __name__ == "__main__":
    import sys
    import pandas as pd
    chrom=sys.argv[1]
    snp = sys.argv[2]
    gene = sys.argv[3]
    molPhen=sys.argv[4]
    
    PhenFile = "/project2/gilad/briana/threeprimeseq/data/molecular_phenos/fastqtl_qqnorm%sphase2.fixed.noChr.txt"%(molPhen)
    GenFile= "/project2/gilad/briana/YRI_geno_hg19/chr%s.dose.filt.vcf"%(chrom)
    outFile = "/project2/gilad/briana/threeprimeseq/data/ApaQTL_examples/qtlSNP_Peak%s%s%s.txt"%(molPhen, snp, gene)
    main(PhenFile, GenFile, outFile, snp, gene,molPhen)
    
```



test this:  


```{bash, eval=F}
python createQTLsnpMolPhenTable.py 3 3:150302010 EIF2A _RNAseq_
```


list for phenos:  

* 4su_30  

* 4su_60  

* RNAseqGeuvadis  

* RNAseq  

* prot  

* ribo

Create a bash script that will use a for loop to run the python script on a all of the phenotypes  

run_createQTLsnpMolPhenTable.sh
```{bash,eval=F}
#!/bin/bash

#SBATCH --job-name=run_createQTLsnpMolPhenTable
#SBATCH --account=pi-yangili1
#SBATCH --time=24:00:00
#SBATCH --output=run_createQTLsnpMolPhenTable.out
#SBATCH --error=run_createQTLsnpMolPhenTable.err
#SBATCH --partition=broadwl
#SBATCH --mem=12G
#SBATCH --mail-type=END

module load python 


chrom=$1
snp=$2
gene=$3

for i in "_4su_30_" "_4su_60_" "_RNAseqGeuvadis_" "_RNAseq_" "_prot." "_ribo_"
do
python createQTLsnpMolPhenTable.py ${chrom} ${snp} ${gene} ${i}
done

```