# Significant negative heritability in LDSC

**URL:** <https://isgw-forum.colorado.edu/t/significant-negative-heritability-in-ldsc/522>\
**Category:** Downstream analysis with genomic data\
**Created:** [March 10, 2023, 4:12am UTC](https://isgw-forum.colorado.edu/t/significant-negative-heritability-in-ldsc/522 "2023-03-10T04:12:58Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Giaco](https://isgw-forum.colorado.edu/letter_avatar_proxy/v4/letter/g/a3d4f5/32.png) [@Giaco](https://isgw-forum.colorado.edu/u/Giaco)\
**Post date:** [March 10, 2023, 4:12am UTC](https://isgw-forum.colorado.edu/t/significant-negative-heritability-in-ldsc/522/1 "2023-03-10T04:12:58Z")

</div>

I wondered whether someone could help me understand negative (and significant) h2snp estimates (estimated via either LDSC or stratified LDSC). And if there is some good reference on how to deal with them.

Many thanks!  
Giaco

---

<div class="post-metadata">

**Author:** ![Andrew\_Grotzinger](https://isgw-forum.colorado.edu/user_avatar/isgw-forum.colorado.edu/andrew_grotzinger/32/114_2.png) [@Andrew\_Grotzinger](https://isgw-forum.colorado.edu/u/Andrew_Grotzinger)\
**Post date:** [March 10, 2023, 4:50pm UTC](https://isgw-forum.colorado.edu/t/significant-negative-heritability-in-ldsc/522/2 "2023-03-10T16:50:12Z")

</div>

Hi Giaco,

Great question! Raymond Walters previously posted an answer to this on the LDSC github:

> <https://github.com/bulik/ldsc/issues/88>
>
> Hi, I am trying to use your software to calculate the genetic correlation betwee…n phenotype1 and phenotype2, but the estimated heritability of phenotype1 is negative, thus I cannot get the genetic correlation between phenotype1 and phenotype2. why the estimated heritality could be negative? and how to deal with this problem? Thank you.
> Another question is whether we can calculate the genetic correlation just according to the defination of genetic correlation in your online methods, it is more simple and does not occur the negative heritability?
> Need your help.

Negative h2SNP estimates are much more common with stratified LDSC due to the greater sampling variation observed for heritability estimates stratified across the functional annotations. I would say if you are seeing a sizable negative h2SNP estimate for LDSC (e.g., \< -.3) then it would also be good to look at a manhattan and/or QQ-plot of the hapmap3 SNPs used for LDSC estimation to make sure nothing stands out as potentially off, such as lots of floating points in the manhattan.

Best,  
Andrew

---

<div class="post-metadata">

**Author:** ![luciadehoyos](https://isgw-forum.colorado.edu/letter_avatar_proxy/v4/letter/l/b2d939/32.png) [@luciadehoyos](https://isgw-forum.colorado.edu/u/luciadehoyos)\
**Post date:** [March 10, 2023, 10:26pm UTC](https://isgw-forum.colorado.edu/t/significant-negative-heritability-in-ldsc/522/3 "2023-03-10T22:26:17Z")

</div>

Hi Andrew,  
I was wondering how can this negative heritability be statistically significant?  
Thanks!  
Lucia