search

DReichLab / EIG

Eigen tools by Nick Patterson and Alkes Price lab
Other
180 stars 60 forks source link

smartpca decimal places #63

Closed brautaz closed 3 years ago

brautaz commented 3 years ago

Hi nick, how do I specify the n of decimal places I want outputted in my final .evec file? The default is 4, I'd like to have 6 decimals for my pca coordinates. Thx in advance!

bumblenick commented 3 years ago

hiprecision: YES in parameter file

Nick

On Fri, Mar 26, 2021 at 2:31 PM brautaz @.***> wrote:

Hi nick, how do I specify the n of decimal places I want outputted in my final .evec file? The default is 4, I'd like to have 6 decimals for my pca coordinates. Thx in advance!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/DReichLab/EIG/issues/63, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEE77B4UTBNGK5FMRZFUW2TTFTHIDANCNFSM4Z34LOBA .

brautaz commented 3 years ago

I put this in the parameter file and my .evec values are still in 4 decimals

bumblenick commented 3 years ago

check your spelling of parameter hiprec: YES will also work Make sure you are using the latest version

On Fri, Mar 26, 2021 at 4:40 PM brautaz @.***> wrote:

I put this in the parameter file and my .evec values are still in 4 decimals

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/DReichLab/EIG/issues/63#issuecomment-808496386, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEE77B32LRA3NMTM4OACEWDTFTWK3ANCNFSM4Z34LOBA .

brautaz commented 3 years ago

THE INPUT PARAMETERS

PARAMETER NAME: VALUE

genotypename: /Users/brautaz/Desktop/1000G/phter/ph2ter.geno snpname: /Users/brautaz/Desktop/1000G/phter/ph2ter.snp indivname: /Users/brautaz/Desktop/1000G/phter/ph2ter.ind evecoutname: /Users/brautaz/Desktop/1000G/phter/ph2ter.evec evaloutname: /Users/brautaz/Desktop/1000G/phter/ph2ter.eval lsqproject: YES numoutevec: 25 numthreads: 1 hiprecision: YES

smartpca version: 16000

norm used

lsqproject used number of samples used: 1375 number of snps used: 80348

Using the newest code: 

            name     PC1     PC2     PC3  ...    PC23    PC24    PC25         population

0 FAM001:ID001 -0.0070 -0.0086 0.0043 ... -0.0018 -0.0133 0.0164 ph2ter 1 116:ARK-59 0.0425 0.0650 0.0220 ... -0.0039 0.0024 0.0200 Armenian_Hemsheni 2 117:ARK-72 0.0047 0.0393 0.0017 ... 0.0032 0.0013 -0.0018 Armenian_Hemsheni

brautaz commented 3 years ago

Maybe there is a hard-coded solution to this?

bumblenick commented 3 years ago

We are about to put a new release of smartpca on github. I specifically tested hiprecision: YES on the new release. So wait a couple of days. and try again.

Nick

On Tue, Mar 30, 2021 at 9:06 AM brautaz @.***> wrote:

Maybe there is a hard-coded solution to this?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/DReichLab/EIG/issues/63#issuecomment-810219317, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEE77B355TVFX7FLNZBMGSTTGHEFDANCNFSM4Z34LOBA .

bumblenick commented 3 years ago

New version of smartpca (version 18140) now on github, For more precision (6 decimal places on evec output, code hiprecision: YES or hiprec: YES

On Tue, Mar 30, 2021 at 9:43 AM Nick Patterson @.***> wrote:

We are about to put a new release of smartpca on github. I specifically tested hiprecision: YES on the new release. So wait a couple of days. and try again.

Nick

On Tue, Mar 30, 2021 at 9:06 AM brautaz @.***> wrote:

Maybe there is a hard-coded solution to this?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/DReichLab/EIG/issues/63#issuecomment-810219317, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEE77B355TVFX7FLNZBMGSTTGHEFDANCNFSM4Z34LOBA .

brautaz commented 3 years ago

This worked! TY

bumblenick commented 3 years ago

Great; I don't know why it didn't work before but I am not going to worry about it.

Nick

On Thu, Apr 1, 2021 at 8:32 PM brautaz @.***> wrote:

This worked! TY

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/DReichLab/EIG/issues/63#issuecomment-812250363, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEE77B5UPJUTJQZDTXCYV7DTGUGCZANCNFSM4Z34LOBA .