mtag
mtag copied to clipboard
JonJala
baselineLF_v2.2.UKB.tar.gz
Hi,
I want to use UKB data as reference(baselineLF_v2.2.UKB/). But I can't find a good way to change the defalut reference(ld_ref_panel). I saw it is wrote in the script. I have no clue to change it. Do you know how can I change it easier? Thank you.
Best regards, Kai
Hello,
Have you tried using the "--ld_ref_panel" flag? (running "mtag.py -h" will give you a breakdown of the different flags and options)
That will allow you to pass in a directory path to override the ld_ref_panel/eur_w_ld_chr/ one.
Hi，
I tried. But I got "MemoryError". I used 120G.
Best, Kai
On Mon, Dec 28, 2020 at 9:56 PM Jonathan Jala [email protected] wrote:
Hello,
Have you tried using the "--ld_ref_panel" flag? (running "mtag.py -h" will give you a breakdown of the different flags and options)
That will allow you to pass in a directory path to override the ld_ref_panel/eur_w_ld_chr/ one.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-751861765, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMSN33NPAU4BUUJSPZS6273SXDWJTANCNFSM4VMMWHGQ .
The panel that comes with MTAG looks to be about 55MB. Is 120G how large the directory you're specifying is? What you're trying to use is over 2000 times the size, in that case.
I saw the reference (ldscore.gz). They are too big and there is no "CM MAF LD". I want to use UK biobank reference. Mtag's reference is 1000 genome. So I downloaded baselineLF_v2.2.UKB.tar.gz from https://alkesgroup.broadinstitute.org/LDSCORE/. This contains 187 annotations for 19,476,620 UK Biobank SNPs with MAF>=0.1%. So I want to try this.
On Tue, Dec 29, 2020 at 5:26 PM Jonathan Jala [email protected] wrote:
The panel that comes with MTAG looks to be about 55MB. Is 120G how large the directory you're specifying is? What you're trying to use is over 2000 times the size, in that case.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752145751, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMSN33MFVQAENZIS6DKCZE3SXH7LRANCNFSM4VMMWHGQ .
Ah ok. You could try removing the annotated columns for what you send into MTAG, since that will reduce the size of things by almost a factor of 200. See if you still run out of memory then?
On Tue, Dec 29, 2020, 4:14 PM Kai6662 [email protected] wrote:
I saw the reference (ldscore.gz). They are too big and there is no "CM MAF LD". I want to use UK biobank reference. Mtag's reference is 1000 genome. So I downloaded baselineLF_v2.2.UKB.tar.gz from https://alkesgroup.broadinstitute.org/LDSCORE/. This contains 187 annotations for 19,476,620 UK Biobank SNPs with MAF>=0.1%. So I want to try this.
On Tue, Dec 29, 2020 at 5:26 PM Jonathan Jala [email protected] wrote:
The panel that comes with MTAG looks to be about 55MB. Is 120G how large the directory you're specifying is? What you're trying to use is over 2000 times the size, in that case.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752145751, or unsubscribe < https://github.com/notifications/unsubscribe-auth/AMSN33MFVQAENZIS6DKCZE3SXH7LRANCNFSM4VMMWHGQ
.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752242935, or unsubscribe https://github.com/notifications/unsubscribe-auth/APIOF56JGMTT4RHRSW4ISR3SXJBCVANCNFSM4VMMWHGQ .
But I checked those files. They don't have MAF, M and L2. Then I found some information in README like "you have access to UK10K in order to compute LD scores (note that we cannot share UK10K LD scores due to UK10K sharing policy)". That means I can't get LD scores? I have to compute based on what I have? Thank you.
On Tue, Dec 29, 2020 at 10:44 PM Jonathan Jala [email protected] wrote:
Ah ok. You could try removing the annotated columns for what you send into MTAG, since that will reduce the size of things by almost a factor of 200. See if you still run out of memory then?
On Tue, Dec 29, 2020, 4:14 PM Kai6662 [email protected] wrote:
I saw the reference (ldscore.gz). They are too big and there is no "CM MAF LD". I want to use UK biobank reference. Mtag's reference is 1000 genome. So I downloaded baselineLF_v2.2.UKB.tar.gz from https://alkesgroup.broadinstitute.org/LDSCORE/. This contains 187 annotations for 19,476,620 UK Biobank SNPs with MAF>=0.1%. So I want to try this.
On Tue, Dec 29, 2020 at 5:26 PM Jonathan Jala [email protected] wrote:
The panel that comes with MTAG looks to be about 55MB. Is 120G how large the directory you're specifying is? What you're trying to use is over 2000 times the size, in that case.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752145751, or unsubscribe <
https://github.com/notifications/unsubscribe-auth/AMSN33MFVQAENZIS6DKCZE3SXH7LRANCNFSM4VMMWHGQ
.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752242935, or unsubscribe < https://github.com/notifications/unsubscribe-auth/APIOF56JGMTT4RHRSW4ISR3SXJBCVANCNFSM4VMMWHGQ
.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752251729, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMSN33NVCZHCVRBCMFAIRH3SXJES3ANCNFSM4VMMWHGQ .
Hi,
I haven't used the Price Lab files you reference above, so I don't think I can help you with that. If you can't trim down those files to have just what you need, you may have to either create your own using some other publicly available reference data or use publicly available LD scores from elsewhere.
Sorry I couldn't be of more help.
Patrick
On Wed, Dec 30, 2020 at 6:07 PM Kai6662 [email protected] wrote:
But I checked those files. They don't have MAF, M and L2. Then I found some information in README like "you have access to UK10K in order to compute LD scores (note that we cannot share UK10K LD scores due to UK10K sharing policy)". That means I can't get LD scores? I have to compute based on what I have? Thank you.
On Tue, Dec 29, 2020 at 10:44 PM Jonathan Jala [email protected] wrote:
Ah ok. You could try removing the annotated columns for what you send into MTAG, since that will reduce the size of things by almost a factor of
See if you still run out of memory then?
On Tue, Dec 29, 2020, 4:14 PM Kai6662 [email protected] wrote:
I saw the reference (ldscore.gz). They are too big and there is no "CM MAF LD". I want to use UK biobank reference. Mtag's reference is 1000 genome. So I downloaded baselineLF_v2.2.UKB.tar.gz from https://alkesgroup.broadinstitute.org/LDSCORE/. This contains 187 annotations for 19,476,620 UK Biobank SNPs with MAF>=0.1%. So I want to try this.
On Tue, Dec 29, 2020 at 5:26 PM Jonathan Jala < [email protected]> wrote:
The panel that comes with MTAG looks to be about 55MB. Is 120G how large the directory you're specifying is? What you're trying to use is over 2000 times the size, in that case.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752145751, or unsubscribe <
https://github.com/notifications/unsubscribe-auth/AMSN33MFVQAENZIS6DKCZE3SXH7LRANCNFSM4VMMWHGQ
.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752242935, or unsubscribe <
https://github.com/notifications/unsubscribe-auth/APIOF56JGMTT4RHRSW4ISR3SXJBCVANCNFSM4VMMWHGQ
.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752251729, or unsubscribe < https://github.com/notifications/unsubscribe-auth/AMSN33NVCZHCVRBCMFAIRH3SXJES3ANCNFSM4VMMWHGQ
.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/122#issuecomment-752784839, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFBUB5IGOH2LJG4WP7GYTJDSXOXDRANCNFSM4VMMWHGQ .
