search

omerwe / polyfun

PolyFun (POLYgenic FUNctionally-informed fine-mapping)
MIT License
86 stars 21 forks source link

estimate the relative contribution of eQTLs to the heritability of a complex trait #30

Closed montenegrina closed 4 years ago

montenegrina commented 4 years ago

Hello,

I would like to estimate the relative contribution of eQTLs to the heritability of a complex trait. Can you please tell em if your software could be used for that purpose. If not can you please recommend any other software that would work with GWAS data?

Cheers, Ana

omerwe commented 4 years ago

Hi, I recommend MESC (Github: https://github.com/douglasyao/mesc)

montenegrina commented 4 years ago

Hi Omer,

MESC seems to be for estimating heritability from expression data. Do you by any chance know about software that does that from GWAS aside LDSC?

Thanks Ana

On Tue, Aug 4, 2020 at 7:51 AM Omer Weissbrod notifications@github.com wrote:

Hi, I recommend MESC https://www.nature.com/articles/s41588-020-0625-2 (Github: https://github.com/douglasyao/mesc)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-668577018, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACF3RTDG2LYLEMJSS63OKXTR6773LANCNFSM4PT4IKBQ .

omerwe commented 4 years ago

Hi, do you mean estimating the h2 explained by a subset of SNPs (that happen to be eQTLs)? Are you interested in eQTLS in a specific locus, or many eQTLs across the entire genome?

If it's the latter, you can use BOLT-REML or S-LDSC. If it's the former, you can get a rough approximation by applying SuSiE and summing the posterior mean squared effect sizes of the eQTLs.

On Tue, Aug 4, 2020, 9:44 AM montenegrina notifications@github.com wrote:

Hi Omer,

MESC seems to be for estimating heritability from expression data. Do you by any chance know about software that does that from GWAS aside LDSC?

Thanks Ana

On Tue, Aug 4, 2020 at 7:51 AM Omer Weissbrod notifications@github.com wrote:

Hi, I recommend MESC https://www.nature.com/articles/s41588-020-0625-2 (Github: https://github.com/douglasyao/mesc)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-668577018, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ACF3RTDG2LYLEMJSS63OKXTR6773LANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-668604880, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACNCB46YDCHSEFY4YU3ISP3R7AGDNANCNFSM4PT4IKBQ .

montenegrina commented 4 years ago

Hi Omer

I am interested in estimating heritability for many EQTLs across the genome not in specific locus. Can you please advise.

Cheers Ana

On Wed, 5 Aug 2020 at 09:58, Omer Weissbrod notifications@github.com wrote:

Hi, do you mean estimating the h2 explained by a subset of SNPs (that happen to be eQTLs)? Are you interested in eQTLS in a specific locus, or many eQTLs across the entire genome?

If it's the latter, you can use BOLT-REML or S-LDSC. If it's the former, you can get a rough approximation by applying SuSiE and summing the posterior mean squared effect sizes of the eQTLs.

On Tue, Aug 4, 2020, 9:44 AM montenegrina notifications@github.com wrote:

Hi Omer,

MESC seems to be for estimating heritability from expression data. Do you by any chance know about software that does that from GWAS aside LDSC?

Thanks Ana

On Tue, Aug 4, 2020 at 7:51 AM Omer Weissbrod notifications@github.com wrote:

Hi, I recommend MESC < https://www.nature.com/articles/s41588-020-0625-2> (Github: https://github.com/douglasyao/mesc)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-668577018, or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACF3RTDG2LYLEMJSS63OKXTR6773LANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-668604880, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ACNCB46YDCHSEFY4YU3ISP3R7AGDNANCNFSM4PT4IKBQ

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669243250, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACF3RTDLWWSJIYXNRMMR5VTR7FXSZANCNFSM4PT4IKBQ .

omerwe commented 4 years ago

Hi Ana,

Any heritability-estimation tool that can perform a multiple-components analysis can do this. Two possible tools are:

  1. BOLT-REML applied with two components: eQTL SNPs and non-eQTL SNPs
  2. S-LDSC applied with two annotations: eQTL SNPs and non-eQTL SNPs

Cheers,

Omer

On Wed, Aug 5, 2020 at 11:46 AM montenegrina notifications@github.com wrote:

Hi Omer

I am interested in estimating heritability for many EQTLs across the genome not in specific locus. Can you please advise.

Cheers Ana

On Wed, 5 Aug 2020 at 09:58, Omer Weissbrod notifications@github.com wrote:

Hi, do you mean estimating the h2 explained by a subset of SNPs (that happen to be eQTLs)? Are you interested in eQTLS in a specific locus, or many eQTLs across the entire genome?

If it's the latter, you can use BOLT-REML or S-LDSC. If it's the former, you can get a rough approximation by applying SuSiE and summing the posterior mean squared effect sizes of the eQTLs.

On Tue, Aug 4, 2020, 9:44 AM montenegrina notifications@github.com wrote:

Hi Omer,

MESC seems to be for estimating heritability from expression data. Do you by any chance know about software that does that from GWAS aside LDSC?

Thanks Ana

On Tue, Aug 4, 2020 at 7:51 AM Omer Weissbrod < notifications@github.com> wrote:

Hi, I recommend MESC < https://www.nature.com/articles/s41588-020-0625-2> (Github: https://github.com/douglasyao/mesc)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <https://github.com/omerwe/polyfun/issues/30#issuecomment-668577018 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACF3RTDG2LYLEMJSS63OKXTR6773LANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-668604880, or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACNCB46YDCHSEFY4YU3ISP3R7AGDNANCNFSM4PT4IKBQ

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669243250, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ACF3RTDLWWSJIYXNRMMR5VTR7FXSZANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669270013, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACNCB4ZBJ2MOSESREAKX4ADR7F5FDANCNFSM4PT4IKBQ .

montenegrina commented 4 years ago

Hi Omer,

Of these two tools only LDSC works with GWAS data? I don’t have genotypes

On Wed, 5 Aug 2020 at 12:24, Omer Weissbrod notifications@github.com wrote:

Hi Ana,

Any heritability-estimation tool that can perform a multiple-components analysis can do this. Two possible tools are:

  1. BOLT-REML applied with two components: eQTL SNPs and non-eQTL SNPs
  2. S-LDSC applied with two annotations: eQTL SNPs and non-eQTL SNPs

Cheers,

Omer

On Wed, Aug 5, 2020 at 11:46 AM montenegrina notifications@github.com wrote:

Hi Omer

I am interested in estimating heritability for many EQTLs across the genome not in specific locus. Can you please advise.

Cheers Ana

On Wed, 5 Aug 2020 at 09:58, Omer Weissbrod notifications@github.com wrote:

Hi, do you mean estimating the h2 explained by a subset of SNPs (that happen to be eQTLs)? Are you interested in eQTLS in a specific locus, or many eQTLs across the entire genome?

If it's the latter, you can use BOLT-REML or S-LDSC. If it's the former, you can get a rough approximation by applying SuSiE and summing the posterior mean squared effect sizes of the eQTLs.

On Tue, Aug 4, 2020, 9:44 AM montenegrina notifications@github.com wrote:

Hi Omer,

MESC seems to be for estimating heritability from expression data. Do you by any chance know about software that does that from GWAS aside LDSC?

Thanks Ana

On Tue, Aug 4, 2020 at 7:51 AM Omer Weissbrod < notifications@github.com> wrote:

Hi, I recommend MESC < https://www.nature.com/articles/s41588-020-0625-2> (Github: https://github.com/douglasyao/mesc)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub < https://github.com/omerwe/polyfun/issues/30#issuecomment-668577018 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACF3RTDG2LYLEMJSS63OKXTR6773LANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <https://github.com/omerwe/polyfun/issues/30#issuecomment-668604880 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACNCB46YDCHSEFY4YU3ISP3R7AGDNANCNFSM4PT4IKBQ

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669243250, or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACF3RTDLWWSJIYXNRMMR5VTR7FXSZANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669270013, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ACNCB4ZBJ2MOSESREAKX4ADR7F5FDANCNFSM4PT4IKBQ

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669323412, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACF3RTDIVZK4NUKDVENRGZTR7GITFANCNFSM4PT4IKBQ .

omerwe commented 4 years ago

Yes, you need to use LDSC if you only have sumstats (or some other tool such as SumHer).

On Wed, Aug 5, 2020 at 2:06 PM montenegrina notifications@github.com wrote:

Hi Omer,

Of these two tools only LDSC works with GWAS data? I don’t have genotypes

On Wed, 5 Aug 2020 at 12:24, Omer Weissbrod notifications@github.com wrote:

Hi Ana,

Any heritability-estimation tool that can perform a multiple-components analysis can do this. Two possible tools are:

  1. BOLT-REML applied with two components: eQTL SNPs and non-eQTL SNPs
  2. S-LDSC applied with two annotations: eQTL SNPs and non-eQTL SNPs

Cheers,

Omer

On Wed, Aug 5, 2020 at 11:46 AM montenegrina notifications@github.com wrote:

Hi Omer

I am interested in estimating heritability for many EQTLs across the genome not in specific locus. Can you please advise.

Cheers Ana

On Wed, 5 Aug 2020 at 09:58, Omer Weissbrod notifications@github.com wrote:

Hi, do you mean estimating the h2 explained by a subset of SNPs (that happen to be eQTLs)? Are you interested in eQTLS in a specific locus, or many eQTLs across the entire genome?

If it's the latter, you can use BOLT-REML or S-LDSC. If it's the former, you can get a rough approximation by applying SuSiE and summing the posterior mean squared effect sizes of the eQTLs.

On Tue, Aug 4, 2020, 9:44 AM montenegrina notifications@github.com wrote:

Hi Omer,

MESC seems to be for estimating heritability from expression data. Do you by any chance know about software that does that from GWAS aside LDSC?

Thanks Ana

On Tue, Aug 4, 2020 at 7:51 AM Omer Weissbrod < notifications@github.com> wrote:

Hi, I recommend MESC < https://www.nature.com/articles/s41588-020-0625-2> (Github: https://github.com/douglasyao/mesc)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub < https://github.com/omerwe/polyfun/issues/30#issuecomment-668577018 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACF3RTDG2LYLEMJSS63OKXTR6773LANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub < https://github.com/omerwe/polyfun/issues/30#issuecomment-668604880 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACNCB46YDCHSEFY4YU3ISP3R7AGDNANCNFSM4PT4IKBQ

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <https://github.com/omerwe/polyfun/issues/30#issuecomment-669243250 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACF3RTDLWWSJIYXNRMMR5VTR7FXSZANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669270013, or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACNCB4ZBJ2MOSESREAKX4ADR7F5FDANCNFSM4PT4IKBQ

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669323412, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ACF3RTDIVZK4NUKDVENRGZTR7GITFANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669347879, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACNCB4ZMOFNPHFQEKQRAWNLR7GNTLANCNFSM4PT4IKBQ .

montenegrina commented 4 years ago

Thank you so much!

On Wed, 5 Aug 2020 at 13:09, Omer Weissbrod notifications@github.com wrote:

Yes, you need to use LDSC if you only have sumstats (or some other tool such as SumHer).

On Wed, Aug 5, 2020 at 2:06 PM montenegrina notifications@github.com wrote:

Hi Omer,

Of these two tools only LDSC works with GWAS data? I don’t have genotypes

On Wed, 5 Aug 2020 at 12:24, Omer Weissbrod notifications@github.com wrote:

Hi Ana,

Any heritability-estimation tool that can perform a multiple-components analysis can do this. Two possible tools are:

  1. BOLT-REML applied with two components: eQTL SNPs and non-eQTL SNPs
  2. S-LDSC applied with two annotations: eQTL SNPs and non-eQTL SNPs

Cheers,

Omer

On Wed, Aug 5, 2020 at 11:46 AM montenegrina <notifications@github.com

wrote:

Hi Omer

I am interested in estimating heritability for many EQTLs across the genome not in specific locus. Can you please advise.

Cheers Ana

On Wed, 5 Aug 2020 at 09:58, Omer Weissbrod < notifications@github.com> wrote:

Hi, do you mean estimating the h2 explained by a subset of SNPs (that happen to be eQTLs)? Are you interested in eQTLS in a specific locus, or many eQTLs across the entire genome?

If it's the latter, you can use BOLT-REML or S-LDSC. If it's the former, you can get a rough approximation by applying SuSiE and summing the posterior mean squared effect sizes of the eQTLs.

On Tue, Aug 4, 2020, 9:44 AM montenegrina < notifications@github.com> wrote:

Hi Omer,

MESC seems to be for estimating heritability from expression data. Do you by any chance know about software that does that from GWAS aside LDSC?

Thanks Ana

On Tue, Aug 4, 2020 at 7:51 AM Omer Weissbrod < notifications@github.com> wrote:

Hi, I recommend MESC < https://www.nature.com/articles/s41588-020-0625-2> (Github: https://github.com/douglasyao/mesc)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub < https://github.com/omerwe/polyfun/issues/30#issuecomment-668577018 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACF3RTDG2LYLEMJSS63OKXTR6773LANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub < https://github.com/omerwe/polyfun/issues/30#issuecomment-668604880 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACNCB46YDCHSEFY4YU3ISP3R7AGDNANCNFSM4PT4IKBQ

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub < https://github.com/omerwe/polyfun/issues/30#issuecomment-669243250 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACF3RTDLWWSJIYXNRMMR5VTR7FXSZANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <https://github.com/omerwe/polyfun/issues/30#issuecomment-669270013 , or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACNCB4ZBJ2MOSESREAKX4ADR7F5FDANCNFSM4PT4IKBQ

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669323412, or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ACF3RTDIVZK4NUKDVENRGZTR7GITFANCNFSM4PT4IKBQ

.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669347879, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ACNCB4ZMOFNPHFQEKQRAWNLR7GNTLANCNFSM4PT4IKBQ

.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/omerwe/polyfun/issues/30#issuecomment-669349522, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACF3RTDQ6FCSTA3KYI27EPLR7GN6XANCNFSM4PT4IKBQ .