Hi Guys,
I have sets of genotype data. The data are more than 99% (presumably) concordant. What would be the best way to represent them with proper statistics and figures? Here in the table below, I need to compare the concordance between the given genotypes (geno1 and geno2, geno3 and geno4, and geno5 and geno6). Here, geno1 and geno2 are 100% concordant, geno3 and geno4 have some genotypes that are not concordant and geno5 and geno6 have genotypes that are least concordant with each other. What would be the best way to analyse this data (assuming I have high school level of statistics knowledge)? Thank you!
geno1 geno2 geno3 geno4 geno5 geno6
0/1 0/1 0/0 0/0 1/1 1/0
0/0 0/0 0/1 0/1 0/0 0/1
0/1 0/1 0/0 0/0 0/1 0/1
0/0 0/0 0/0 0/0 1/1 1/1
0/0 0/0 0/1 0/1 0/0 0/0
0/1 0/1 0/1 0/0 0/1 0/0
1 answer
The easiest way to do will be as follow:
Assuming your information were stored in a text file called test.txt
Then you can:
data = read.table("test.txt", header=T)
concordant = matrix(NA, ncol(data), ncol(data))
colnames(concordant) = colnames(data)
rownames(concordant) = rownames(data)
for(i in 1:ncol(data)){
for(j in i:ncol(data)){
concordant[i,j] = sum(as.character(data[,i])==as.character(data[,j]))/ncol(data)
}}
If you have a large table and don't want to waste time by computing more of the concordance information, you can change the line
result = data.frame(A=character(ncol(data)/2), B=character(ncol(data)/2), Concordance=numeric(ncol(data)/2),stringsAsFactors=FALSE);
j = 1;
for(i in seq(1, ncol(data), 2)){
result[j,1] = colnames(data)[i];
result[j,2] = colnames(data)[i+1];
result[j,3] = sum(as.character(data[,i])==as.character(data[,i+1]))/ncol(data)
j= j+1;
}
However, if you consider 0/1 and 1/0 as the same, then you simple just change all the instance of 1/0 to 0/1 or the other way round
One quick question: How do I skip any column in one of the two column-sets if they are empty (as shown in table below)? I want to exclude them from the calculation. Thank you again for your help!
geno1 geno2 geno3 geno4 geno5 geno6
0/1 0/1 0/0 0/0 1/1 1/0
0/0 0/0 0/1 0/0 0/1
0/1 0/1 0/0 0/0 0/1 0/1
0/0 0/0 0/0 1/1
0/0 0/0 0/1 0/1 0/0 0/0
0/1 0/1 0/1 0/0 0/1 0/0
Assuming that your separator is tab, and empty cells are represented as " " or ""
result = data.frame(A=character(ncol(data)/2), B=character(ncol(data)/2), Concordance=numeric(ncol(data)/2),stringsAsFactors=FALSE);
j = 1;
for(i in seq(1, ncol(data), 2)){
result[j,1] = colnames(data)[i];
result[j,2] = colnames(data)[i+1];
result[j,3] = sum(as.character(data[,i])==as.character(data[,i+1]))/sum(!((data[,i] == " " | data[,i] == "") | (data[,i+1]==" " | data[,i] == "")))
j= j+1;
}
Log in to answer this question.
-
single cell data integration approach
written by Jay •
-
Logistic Regression in R
written by Onurb1983 •
-
Strange order of multi-allelic site in VCF file
written by miaowzai
-
Extracting genotypes and filtering from a VCF file containing multiple exomes of families
written by gradstudentNew •
-
phasing SNPs with 3 siblings
written by firestar
-
Why does GATK LeftAlignAndTrimVariants set a missing genotype to 0/0?
written by Apprentice
-
Statistical analysis to compare three different variables
written by MAPK
-
R programming: genotype concordance
written by MAPK
-
R programming: compare columns to column and get the mismatch
written by MAPK
-
R programming question: genotyping data table manipulation2
written by MAPK
What's your purpose? Are you trying to compare whether if gene1 and gene2 are significantly more concordant than say, gene1 and gene5?
Or are you trying to get a concordant table? If so you just need to represent it by an n by n table:
Yes, I am trying to compare whether gene1 and gene2 are significantly more concordant than say, gene5 and gene6. Basically I want to show the concordance level of all pair set in graphical form. Concordance between pair sets of geno1 and geno2, geno 3 and geno4 and geno5 and geno6.
Do you consider 0/1 and 1/0 as concordant?
No, it is not concordant. Thank you!