search

SATVILab / PipelineAnalysisACS

Analysis functions for pipeline for ACS-related analyses.
Other
0 stars 0 forks source link

Assess whether the freq_bs change in `targets` branch was correct #11

Closed MiguelRodo closed 2 years ago

MiguelRodo commented 2 years ago

Related to #7

MiguelRodo commented 2 years ago

In PipelineAnalysis@targets, we changed this

dplyr::rename(freq_bs = prop_bs) %>%

to this

dplyr::mutate(freq_bs = prop_bs * 1e2) %>%

Looking at the code of the function this belongs to, .subtract_background, prop_bs is instead a proportion and not a frequency. So this is correct, ito the column names anyway.

However, when I look at PipelineAnalysisACS@main, I see that this column freq_bs is treated as a proportion rather than a frequency. So despite that it's called a frequency, we treat it as a proportion (which it was).

The function itself is only called once there, in .prep_dr_cytokines_freq in lines 518-519. The resultant column, freq_bs, is then treated as a proportion in line 544. And other than that, there is no reference to freq_bs again. So, I'm not sure if this is why freq_bs became an issue when plotting, but clearly it's not a bug for data processing (as we would expect, as the fitted values and raw values for the frequencies look reasonable).

So we should not take this change.

I think in AnalysisACSCyTOFNKBCells, they don't actually use the .prep_dr_cytokines_freq function, so they would need the change as presumably they treat freq_bs as a frequency.