To keep the model of minimum p-value from the file

Question

[khn]

Hello

My file lists a millions of SNPs with their p-values by additive, dominant, and recessive models. I want to keep the SNPs with a model for which the p-value is the lowest.

Original file.txt

ID,y,z,P-value
rs1,a,b,0.22
rs1,a,b,0.35
rs1,a,b,0.45
rs2,c,d,0.06
rs2,c,d,0.20
rs2,c,d,0.46
rs3,e,f,0.002
rs3,e,f,0.98
rs3,e,f,1.0

The file.txt that I want is...

ID,y,z,P-value
rs1,a,b,0.22
rs2,c,d,0.06
rs3,e,f,0.002

How to do this? Thanks!

[German.M.Demidov]

since the property of p-value is uniformness under the null hypothesis and it will be totally broken by chosin the min from 3 models, I worn you not to do so - you won't have any basis to do the further analysis after such selection

[khn]

Thank you for your comment. I just wanted to draw a manhattan plot based on this. But I think I should just show three manhattan plots, based on the model.

[German.M.Demidov]

Yeap, Manhattan plot would not make any sense if you choose the min =/ 3 plots sounds good!

[zx8754]

This type of questions are more suitable for StackOverflow, see:

• unix How to select rows with minimum value in each group based on first column as ID?
• R Extract the maximum value within each group in a dataframe

Answer 1

[khn]

tail -n +2 original_file > txt sort -t, -k 4g txt | awk -F, '!visited[$1]++' | sort -k2,2 -k3,3 >> final_file

[zx8754]

Please format your code, and add some description on what each step is doing.