Stefano Mangiola

What happens if I scale twice? What should we do. - do it again? - through a warning?

I found that possibly regressing here is not the answer, but discarting some dimension might be. https://github.com/welch-lab/liger/issues/113

Hello @antass, you have only one value of treatment. ```r > my_mat |> distinct(treatment) # A tibble: 1 × 1 treatment 1 AG120 ``` It is a bit strange to...

At the moment no, is all automatic. I see now your use case make sense. Let me ponder on the cost of adding a new argument. Let's leave this issue...

( Just to second, recently Rstudio and stan became quite incompatible. If I am debugging a function and create a stan model (of save) from a file more than 3...

The on-the-fly fix is ```r fit@stan_args[[1]]$method = "hmc" ```

I was looking for something like this ``` rstan::summary(fit, c("y_hat", "overdispersion"))$summary %>% as_tibble(rownames="parameter") %>% separate(parameter, c("term", "tube", "gene"), sep = "[\\[\\],]") ```

I was thinking, would be otherwise (maybe even better) possible to parallelise all *_samples functions with clever partitioning? For example ```R happiness = fit3Gb %>% gather_samples(beta[a,b], cores = 40) ```

I understand that new database connections should not be created for every query, but since many people do, is there a solution to that? It seems that once the error...

I agree it is not easy to reproduce (I did a quick stress test and nothing), but I bet that if you stop saving connection in a variable and pipe...