[Susie] ValueError: prior variance must be non-negative

[lintingyi2014]

please help me fix this

SuSiE fine-mapping

• 734 samples
• 22156 phenotypes
• 69 covariates
• 8526609 variants
• cis-window: ±1,000,000
• applying in-sample MAF >= 0.01 filter
• checking phenotypes: 22156/22156 ** dropping 37 phenotypes without variants in cis-window
• fine-mapping processing phenotype 8/22119Traceback (most recent call last): File "", line 1, in File "/usr/local/lib/python3.10/dist-packages/tensorqtl/susie.py", line 705, in map res = susie(genotypes_res_t.T, phenotype_res_t.T, L=L, File "/usr/local/lib/python3.10/dist-packages/tensorqtl/susie.py", line 495, in susie s = init_finalize(s) File "/usr/local/lib/python3.10/dist-packages/tensorqtl/susie.py", line 99, in init_finalize raise ValueError("prior variance must be non-negative") ValueError: prior variance must be non-negative

[francois-a]

Can you please provide an example dataset that reproduces this issue?

[lintingyi2014]

May I send it to you in private?

[francois-a]

Sure, no problem.

[lintingyi2014]

sorry was a problem on my end! It is solved now.

Thanks!