[ X ] Create function that returns a dict with all the activities, aux vars, behavior, animals for a given proj_meta file
[ X ] Clean up data * [ X ] Remove duplicate grating onsets (keep first) * [ X ] Function to pad n-d array to desired dims * [ ] Make it faster with fewer allocations * [ X ] Remove "flips" from the condition 3 recordings where we used 100% flip occurence to show the right grating * Later * [ ] Remove duplicate grating onsets (keep longest) * [ ] Add option to keep incomplete traversals
[ ] Reproduce paper figures * [ ] Figure 1 * [ X ] Classification of grating + position across conditions * [ ] // // across timepoints (/ hours?) * [ ] Plot some cell examples (maybe extract feature weights from random forest?) * [ ] Figure 2 * [ X ] Spatial binning of activity * [ ] Show mean grating-evoked activity per condition * [ ] Show distribution of peaks for selective cells * [ ] Show (clustered?) activity in spatial coordinates * [ ] Figure 3 * [ ] Quantify "grating belief" as a function of time/condition * [ ] Plot anticipatory vs visual cell activity before/after flips (and other grating onsets). Find a sensible way to do it for cell pairs
* [ ] Figure 4
* [ ] Plot population response to omission
* [ ] Plot "omission-selective" cells
[ ] Other analyses * [ ] What happens with mixed-selectivity cells during flips/omissions? * [ ] Plot peak flip response as function of cell's peak location for A, B * [ ] Predict some behavioral feature from neuronal activity * [ ] Same for CA1, ACC * [ ] Think of interesting analyses to do. Prioritize insight over technical sexiness
[ ] Things to fix / change
* [ ] Undo get_act_dict, use proj_meta directly for analyses
* [ ]
[ ] ?
[ ] Profit