mbk0asis

dta <- read.csv("file:///E:/LAB_DATA/00-LabData/Lab/00--Archive/TCGA_DNA_met/27k_Set2/liver/liver.set2.boxplot..csv",
                header = T)
dta <- read.csv("file:///E:/LAB_DATA/00-LabData/Lab/00--Archive/TCGA_DNA_met/423CpGs_set12357/423CpGs.boxplot.2.csv", header = T)
dim(dta)

##################
types <- data.frame(c(rep("breast.C",793),rep("breast.N",97),
                   rep("colon.C",313),rep("colon.N",38),
                   rep("kidney.C",324),rep("kidney.N",160),
                   rep("liver.C",377),rep("liver.N",50),

mbk0asis

# USING "ggplot2" and "for loop"

types <- data.frame(c(rep("breast.C",793),rep("breast.N",97),
                   rep("colon.C",313),rep("colon.N",38),
                   rep("kidney.C",324),rep("kidney.N",160),
                   rep("liver.C",377),rep("liver.N",50),
                   rep("lung.C",473),rep("lung.N",32),
                   rep("prostate.C",502),rep("prostate.N",50)))
colnames(types) <- "types"

mbk0asis

# Ubuntu 16.04
# R-3.4.1


1. Download R source package

2. Install dependencies
   $  sudo apt-get install fort77 xorg-dev liblzma-dev  libblas-dev gfortran gcc-multilib gobjc++ aptitude libreadline-dev

3. Install 'open-blas'

mbk0asis

library(WGCNA)
allowWGCNAThreads()

#################################################################################
# Loading expression data

setwd('/home/bio0/00-NGS/TCGA_met/data')

set1 <- read.csv("600samples.cosmic170k.csv", header = T, row.names = 1)
all.set <- read.csv("2400samples.27k.set2.2.csv", header = T, row.names = 1)

mbk0asis

library(limma)
library(gplots) 
library(FactoMineR)

##########################################
# Data importing
setwd("D:/00-LabData/Lab/Experiments/2017-01/EpiDriver/T_cell/")
dta<-read.csv("EpiDriver_Tcells_biasCorrection_Final_StdQt.csv", row.names = 1, header = T)

mbk0asis

102_103	insC		4
102	delC		2
101_102	delCC		2
102	C>G		1
101	C>T		4
98_99	insT		1
98	delC		2
98	delC		2
98	C>T		2
97	delT		2

mbk0asis

grep -Po 'c..*' COSMIC_TP53_MUT | sed 's/;/\t/g' | cut -f 1,3 | sed 's/CNT=//g'| grep -Po 'c.\d*(\_?\-?\+?)\d*' > position

grep -Po 'c..*' COSMIC_TP53_MUT | sed 's/;/\t/g' | cut -f 1,3 | sed 's/CNT=//g'| grep -Po '[a-zA-Z]*\>?[a-zA-Z]*\t' > sequence

grep -Po 'c..*' COSMIC_TP53_MUT | sed 's/;/\t/g' | cut -f 3 | sed 's/CNT=//g' > count

paste position sequence count | sed 's/c.//g' > COSMIC_TP53_MUT_all

mbk0asis

17	7676265	COSM45241	AA	A	.	.	GENE=TP53;STRAND=-;CDS=c.103delT;AA=p.L35fs*9;CNT=1
17	7676266	COSM96588	A	AG	.	.	GENE=TP53;STRAND=-;CDS=c.102_103insC;AA=p.P36fs*7;CNT=4
17	7676266	COSM2745164	AG	A	.	.	GENE=TP53;STRAND=-;CDS=c.102delC;AA=p.L35fs*9;CNT=2
17	7676266	COSM48976	AGG	A	.	.	GENE=TP53;STRAND=-;CDS=c.101_102delCC;AA=p.P34fs*8;CNT=2
17	7676267	COSM5387432	G	C	.	.	GENE=TP53;STRAND=-;CDS=c.102C>G;AA=p.P34P;CNT=1
17	7676268	COSM43672	G	A	.	.	GENE=TP53;STRAND=-;CDS=c.101C>T;AA=p.P34L;CNT=4
17	7676270	COSM4765600	G	GA	.	.	GENE=TP53;STRAND=-;CDS=c.98_99insT;AA=p.P36fs*7;CNT=1
17	7676270	COSM1172527	GG	G	.	.	GENE=TP53;STRAND=-;CDS=c.98delC;AA=p.L35fs*9;CNT=2
17	7676270	COSM22928	GG	G	.	.	GENE=TP53;STRAND=-;CDS=c.98delC;AA=p.L35fs*9;CNT=2
17	7676271	COSM4272154	G	A	.	.	GENE=TP53;STRAND=-;CDS=c.98C>T;AA=p.S33F;CNT=2

mbk0asis

#!/bin/bash

export LC_ALL=C

printf "*** SNVcounter\n
	*** 'samtools, bcftools, and bedtools' must be installed to run this script\n
	*** Prepare 'reference fasta' and 'SNV list' in BED format\n
	*** merge consecutive SNVs using\n
	*** 'sort -k1,1 -k2,2n SNV.bed | mergeBed -c 4 -o collapse'\n
	*** usage : ./SNVcounter Input.Bam Reference.Fasta mergedSNV.bed\n\n"

mbk0asis

#!/bin/bash

cat $1 $2 | muscle |fasta_formatter | paste - - - - | cut -f 1,2,4 | sed 's/>M.//g' 
cat $1 $2 | muscle -clw | grep "*" | sed 's/^ *//g;s/ /!/g' | tr '\n' ' ' | sed 's/ //g' | grep -bo "!" | tr '\n' ' '
printf "\n"