Question about TWAS inflation-corrected model

[QiongZhao]

Dear Developers,

Thank you for creating this excellent software. I'm attempting to implement the TWAS inflation-corrected model using my own data, following the PredictDB-Tutorial. However, I noticed that the models I built lack the phi statistic. Can I only use pre-built models from PredictDB (https://predictdb.org/) with this? Or is there a way to build my own models that include phi? Thanks a lot.

[hakyim]

Festus Could you respond? Thanks Haky

…

On Tue, Jun 24, 2025 at 9:39 PM QiongZhao ***@***.***> wrote: *QiongZhao* created an issue (hakyimlab/MetaXcan#215) <#215> Dear Developers, Thank you for creating this excellent software. I'm attempting to implement the TWAS inflation-corrected model using my own data, following the PredictDB-Tutorial <https://github.com/hakyimlab/PredictDB-Tutorial>. However, I noticed that the models I built lack the phi statistic. Can I only use pre-built models from PredictDB (https://predictdb.org/) with this? Or is there a way to build my own models that include phi? Thanks a lot. — Reply to this email directly, view it on GitHub <#215>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAW2ROP7NQBTORYLBC7GZXL3FIDVTAVCNFSM6AAAAACAB6ZFWGVHI2DSMVQWIX3LMV43ASLTON2WKOZTGE3TGOBXHAYTGMY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[Fnyasimi]

@QiongZhao

You can calculate phi for your own models. You need an external genotype where you can use your models to do predictions and simulate null polygenic phenotypes across many simulations. You can use this polygenic_null_sims.R script to perform simulations and obtain the $Z^2$ values for atleast 1000 simulations.

For each gene the average($Z^2$) across the simulations.

Repeat the simulations varying the number of individuals in the genotype (N) and heritability of the null polygenic trait ($h^2$) in the simulation.

For each gene fit a linear model $EZ^2 = \alpha + \Phi*Nh^2 + \epsilon$.

Finally for each gene the $\Phi$ is calculate as $\hat{\Phi}(gene) = \hat{\Phi} + se(\hat{\Phi})$

The tutorial will be out soon.