Florian Privé

Results 380 comments of Florian Privé

It would be nice to get the artifacts produced so that it is clearer where the error comes from. Thanks.

There is always a link to it. But most of the times it results in a 404 error. E.g. https://artifacts.r-hub.io/bigsnpr_1.2.0.tar.gz-db3d8049514647b5948deef51d2fb55b.

Yes, but other builds as well, e.g. Solaris, Windows, CentOS.

You could probably derive these correlations yourself using the formula from `p2cor()`: https://github.com/tshmak/lassosum/blob/master/R/p2cor.R#L20 where `t` is basically `effectsize / se` and `n` is the sample size.

Restart and reinstall?

What I understand from the question: - `v$beta.pgs` is already the "adjusted betas" (meaning directly usable on 0/1/2 genotypes?) - use these effects to compute individual scores with PLINK -...

Seems related to https://github.com/tshmak/lassosum/issues/17.

Thanks for your answer. Memory-mapping basically access data in RAM, as long as you have enough RAM. So, the first access is from disk, but the other accesses are made...

I would need this too (reading only). Any idea how to easily implement this on top of 'mio' for this int-double struct?

What I have currently: ``` struct indval { int i; double x; }; std::error_code error; mio::mmap_source ro_mmap; ro_mmap.map(path, error); const indval * data = reinterpret_cast(ro_mmap.data()); cout i