notes from initial design meeting 20250925

[cmatKhan]

Michael's idea on conceptualizing the underlying data:

universe of possible profiles. A profile is the set of features that uniquely identifies every TF experiment's data. There are possibly mulitple views of this (eg if promoterset is a feature, then the same experiment might be quantified over multiple promoter sets). This actually does describe all of the individual units

Typical user usage pattern:

1. Selectable features, right now just TFs, eg CBF1, or a set (CBF1, RGT1). Rank response, DTO, ... , organism (will eventually have crypto). Dataset (divided into two categories, binding and perturbation)

2. Active dataset (active data set populate the set of active features. If you select kemmeren and mcisaac, then b/c mcisaac has timepoint, then that becomes an active feature that for kemmeren is NA)

3. Active features -- if you select CBF1, then CBF1 exists in some set of datasets. Those datasets become "active", and that populates the "active features" according to those datasets
   I think that for example rank response is a feature and it should only be available if you have selected valid active dataset (at least 1 binding, at least 1 perturbation response). If that is the case, then it makes setting a threshold on rank response is an option