0-99 a: screen -r 65455 anapop
0-99 a: screen -r 71910 plots
100-199 b: screen -r 71925
200-299 c: screen -r 39161
300-399 d: screen -r 39178
400-499 e: screen -r 39189
500-599 f: screen -r 39206
600-639 g: screen -r 39218
nice -19 /system/apps/biosoft/R-2.14.0/bin/R
source("/seppdata/sepp/linkage/release/split/ana_ancest.R")
source("/seppdata/sepp/linkage/release/split/ana_popratio_all.R")
source("/seppdata/sepp/linkage/release/split/dups.R")
source("/seppdata/sepp/linkage/release/split/extract_pos_all.R")
nice -19 /system/apps/biosoft/R-2.14.0/bin/R
source("/seppdata/sepp/linkage/release/split/ana_denisova.R")
source("/seppdata/sepp/linkage/release/split/ana_neander.R")
####################################################################
vcf <- read.table("/seppdata/sepp/linkage/neanderthal_denisova/T_hg19_1000g.Genotypes.vcf",sep="\t")
hg19 <- read.table("/seppdata/sepp/linkage/neanderthal_denisova/hg191000.bed",sep="\t")
hg19M <- paste(as.character(hg19[,1]),as.character(as.numeric(hg19[,2])),sep="")
vcfM <- paste("chr",as.character(vcf[,1]),as.character(as.numeric(vcf[,2])),sep="")
mm <- match(vcfM,hg19M,nomatch=0)
ma <- which(mm>0)
adds <- vcf[ma,5]
refs <- vcf[ma,4]
pos <- hg19[mm[ma],4]
l <- length(hg19[,4])
ad1 <- rep("-",l)
ref1 <- rep("-",l)
ad1[mm[ma]] <- as.character(adds)
ref1[mm[ma]] <- as.character(refs)
denisovaHg19 <- cbind(hg19,ref1,ad1)
denisovaHg19[1:100,]
save(denisovaHg19,file="denisovaHg19.Rda")
> table(denisovaHg19$ad1)
- A,C,G A,C,T A,C,X A,G,C A,G,T A,G,X A,T,C
169327 1 9 2131 5 2 1074 5
A,T,G A,T,X A,X C,A,G C,A,T C,A,X C,G,A C,G,T
9 5220 1198757 10 10 5336 11 5
C,G,X C,T,A C,T,G C,T,X C,X G,A,C G,A,T G,A,X
1489 15 8 3424 807254 5 17 3673
G,C,A G,C,T G,C,X G,T,A G,T,C G,T,X G,X T,A,C
6 3 1700 11 14 5032 808121 7
T,A,G T,A,X T,C,A T,C,G T,C,X T,G,A T,G,C T,G,X
12 5752 8 2 1162 10 7 2355
T,X X
1199518 36442515
chr1 <- which(denisovaHg19$V1=="chr1")
length(chr1)
[1] 3201157
>
denisovaHg19Chr1 <- denisovaHg19[chr1,]
save(denisovaHg19Chr1,file="denisovaHg19Chr1.Rda")
ta <- table(denisovaHg19Chr1$ad1)
ta
- A,C,G A,C,T A,C,X A,G,C A,G,T A,G,X A,T,C A,T,G A,T,X
10632 0 0 200 1 0 85 0 1 404
A,X C,A,G C,A,T C,A,X C,G,A C,G,T C,G,X C,T,A C,T,G C,T,X
93586 0 1 414 1 0 131 0 0 287
C,X G,A,C G,A,T G,A,X G,C,A G,C,T G,C,X G,T,A G,T,C G,T,X
63639 0 3 276 0 0 153 0 0 390
G,X T,A,C T,A,G T,A,X T,C,A T,C,G T,C,X T,G,A T,G,C T,G,X
63833 3 0 429 1 1 79 0 1 179
T,X X
93921 2872506
####################################################################
hg18 <- read.table("/seppdata/sepp/linkage/neanderthal_denisova/hg181000.bed",sep="\t")
hg19 <- read.table("/seppdata/sepp/linkage/neanderthal_denisova/hg191000.bed",sep="\t")
vcf <- read.table("/seppdata/sepp/linkage/neanderthal_denisova/NeandertalGenotypes.vcf",sep="\t")
hg18M <- paste(as.character(hg18[,1]),as.character(as.numeric(hg18[,2])),sep="")
vcfM <- paste(as.character(vcf[,1]),as.character(as.numeric(vcf[,2])),sep="")
mm <- match(vcfM,hg18M,nomatch=0)
ma <- which(mm>0)
adds <- vcf[ma,5]
refs <- vcf[ma,4]
pos <- hg18[mm[ma],4]
mp <- match(pos,hg19[,4],nomatch=0)
mpa <- which(mp>0)
adds <- adds[mpa]
refs <- refs[mpa]
mp <- mp[mpa]
l <- length(hg19[,4])
ad1 <- rep("-",l)
ref1 <- rep("-",l)
ad1[mp] <- as.character(adds)
ref1[mp] <- as.character(refs)
neanderHg19 <- cbind(hg19,ref1,ad1)
neanderHg19[1:100,]
save(neanderHg19,file="neanderHg19.Rda")
> levels(neanderHg19$ad1)
[1] "-" "A,C,G" "A,C,T" "A,C,X" "A,G,C" "A,G,T" "A,G,X" "A,T,C" "A,T,G"
[10] "A,T,X" "A,X" "C,A,G" "C,A,T" "C,A,X" "C,G,A" "C,G,T" "C,G,X" "C,T,A"
[19] "C,T,G" "C,T,X" "C,X" "G,A,C" "G,A,T" "G,A,X" "G,C,A" "G,C,T" "G,C,X"
[28] "G,T,A" "G,T,C" "G,T,X" "G,X" "T,A,C" "T,A,G" "T,A,X" "T,C,A" "T,C,G"
[37] "T,C,X" "T,G,A" "T,G,C" "T,G,X" "T,X" "X"
chr1 <- which(neanderHg19$V1=="chr1")
length(chr1)
neanderHg19Chr1 <- neanderHg19[chr1,]
save(neanderHg19Chr1,file="neanderHg19Chr1.Rda")
ta <- table(neanderHg19Chr1$ad1)
ta
- A,C,G A,C,T A,C,X A,G,C A,G,T A,G,X A,T,C A,T,G A,T,X
1069131 0 0 81 1 0 51 1 0 98
A,X C,A,G C,A,T C,A,X C,G,A C,G,T C,G,X C,T,A C,T,G C,T,X
53106 0 2 99 0 0 44 1 1 171
C,X G,A,C G,A,T G,A,X G,C,A G,C,T G,C,X G,T,A G,T,C G,T,X
36745 0 0 182 0 0 48 0 1 92
G,X T,A,C T,A,G T,A,X T,C,A T,C,G T,C,X T,G,A T,G,C T,G,X
36989 0 1 125 1 0 75 0 0 101
T,X X
54109 1949901
sum(ta)
sum(ta[2:41])
> sum(ta)
[1] 3201157
>
> sum(ta[2:41])
[1] 182125
data <- read.table("../ALL.chr1.merged_beagle_mach.20101123.snps_indels_svs.genotypes_annot.txt",skip=2)
length(data[[10]])
[1] 3201157
>
a1 <- which(data[[10]]<0.05)
length(a1)
[1] 2596777
neanderHg19Chr1Rare <- neanderHg19[a1,]
tar <- table(neanderHg19Chr1Rare$ad1)
tar
sum(tar)
sum(tar[2:41])
> tar
- A,C,G A,C,T A,C,X A,G,C A,G,T A,G,X A,T,C A,T,G A,T,X
853706 0 0 32 0 0 13 0 0 59
A,X C,A,G C,A,T C,A,X C,G,A C,G,T C,G,X C,T,A C,T,G C,T,X
27968 0 1 44 0 0 14 1 0 46
C,X G,A,C G,A,T G,A,X G,C,A G,C,T G,C,X G,T,A G,T,C G,T,X
10106 0 0 48 0 0 18 0 1 36
G,X T,A,C T,A,G T,A,X T,C,A T,C,G T,C,X T,G,A T,G,C T,G,X
10238 0 0 89 1 0 30 0 0 58
T,X X
28594 1665674
>
> sum(tar)
[1] 2596777
ratio rare: 3201157
2596777/3201157
[1] 0.8111995
>
> sum(tar[2:41])
[1] 77397
>
neandertal snps all: 182125
neandertal snps rare: 77397
ratio rare neandertal diffs: 77397/182125
> 77397/182125
[1] 0.4249664