Fastphase cluster determination

Question

[dp0b]

Hi,

I want to determine the number of haplotype clusters there are in my dataset using fastphase. The results are meant to produce a _kselect file, but this never happens and fastphase always uses a K15 for the model.

This is what I have been running- I wanted to test the number of clusters from 5 to 25 going up intervals of 4

./fastPHASE -T10 -ochr26 -KL5 -KU25 -KI5 -KS5 -KM500 -KP.1 plink.chr-26.recode.phase.inp

Am I missing something?

TIA

Answer 1

[Gennady Khvorykh]

Regarding K selection procedure described in the manual (fastPHASE 1.4), I have contacted Paul Scheet, a developer of fastPHASE. He replied that current version of the tool (fastPHASE 1.4.8) doesn't have this option more.

I write wrapper script for fastPHASE to estimate the best K. Feel free to contact me, if you are interested in testing the script for your data.

Answer 2

[Gennady Khvorykh]

Here a workflow I created to estimate K in fastPHASE. Everybody is welcome to explore it.

Answer 3

[bgutg]

Hi! I have the same problem. Did you get the solution?

My comand was this and I don´t get any _kselect file.

./fastPHASE -T20 -C30 -KL6 -KU30 -Ki2 -Ks10 -Km500 -Kp.05 -omerge_Churras280_Merino236 Chu278Mer236qc2simple.inp

Do you need to do it for a single chromosome?

Could you give me a hint? Regards, Beatriz

Answer 4

[call]

Hello,

I have the same problem too. Some solution?

I also tried with the manual example (-KL6 -KU12 -Ki2 -Ks50 -Km1000 -Kp.05), and always uses a K15 for the model

Thanks!