Load ’em up
library(dplyr)
library(ggplot2)
library(magrittr)
library(cowplot)
library(tidyr)
Download subgenome coordinates from gbrowse track at maizegdb.
Download 1cM genetic map from panzea.
Both are on AGPv2
Pull out useful data from genetic map:cut -f 3-5 NAM_phasedImputed_1cM_AllZeaGBSv2.3_allChrs.hmp.txt > ~/Desktop/map.txt
Rewrite as recombination rate and start-stop:
map<-read.table("~/Desktop/map.txt",header=T)
for(chrom in 1:10){
bob=subset(map$pos,map[,1]==chrom)
start=bob[1:(length(bob)-1)]+1
end=bob[2:(length(bob))]
rec=1/diff(bob)
y=data.frame(chrom,start,end,rec)
write.table(y,file="rec.txt",sep="\t",append=T,quote=F,col.names=F,row.names=F)
}
Bedtools FTW:bedtools intersect -a rec.txt -b Annotations.gff3.txt -wo > subgenome_rec.txt
Shameful perl:perl -e 'open FILE, "<subgenome_rec.txt"; while(<FILE>){ $_=~s/Name=\%22//; $_=~s/\%20Ortholog\%20of\%20/\t/g; $_=~s/\%\S+//g; print $_; }' |
cut -f 1-4,13-15 > subgenome_rec_for_analysis.txt
Get data, keep rec segments with perfect overlap:
final_data<-read.table("~/Desktop/subgenome_rec_for_analysis.txt") %>%
set_names(c("chr","start","end","rec","sub","sb","overlap")) %>%
mutate(dist=end-start,rec=rec*1e6,sub=gsub("Maize","M",sub),sb=gsub("Sb","S",sb),inter=interaction(as.factor(sub),as.factor(sb))) %>%
filter(overlap-dist==0)
Take a peek:
Maize1 seems smaller?! Check by ancestral chromosome:
Varies a lot. Let’s check numbers:
summarize(group_by(final_data,sub),mean(rec))
## Source: local data frame [2 x 2]
##
## sub mean(rec)
## 1 M1 2.146173
## 2 M2 2.398824
wilcox.test(subset(final_data$rec,final_data$sub=="M1"),subset(final_data$rec,final_data$sub=="M2"))
##
## Wilcoxon rank sum test with continuity correction
##
## data: subset(final_data$rec, final_data$sub == "M1") and subset(final_data$rec, final_data$sub == "M2")
## W = 190032.5, p-value = 0.002931
## alternative hypothesis: true location shift is not equal to 0
Are maize chromosomes correlated?
## M2
## M1 0.2736872
Not much. Are ancestral chromosomes correlated?
## M2
## M1 -0.06501275
Even less. So local recombination has changed a lot since tetraploidy. And maize1 seems to have \(\approx 10\%\) lower rec than maize1.