Data for reproducing Webster figures

# Data for reproducing Webster figures
This zip file contains the file structure and input data for plugging into the Webster figure generation code repo (https://github.com/joshbiology/gene_fn).
For flat files and Excel spreadsheets containing the actual output of Webster, see our other Figshare (10.6084/m9.figshare.14963561).
This folder is strucured as follows:
```data├── interim│ ├── biomarker│ └── matlab│ ├── depmap_deep│ ├── depmap_denoise│ ├── depmap_grid│ ├── depmap_wide│ ├── durocher│ ├── durocher_no_graph│ └── synthetic└── raw ├── depmap ├── durocher ├── genesets ├── hart ├── humancellmap ├── nusiance_genesets └── prism```

## raw
This folder contains tables from other resources, described below.
* depmap
https://depmap.org/
* durocher
https://pubmed.ncbi.nlm.nih.gov/32649862/
* genesets
Custom genesets from:
* * STAGA/ATAC complexes (Spedale et al., 2012)* * SWI/SNF complexes (Mashtalir et al., 2018).* * Mediator complex (Tsai et al., 2014). * * Integrator complex (Pfleiderer and Galej, 2021; Sabath et al., 2020; Tilley et al., 2021; Zheng et al., 2020)

* hart
Common essential genesets from Hart et al. 2015
* humancellmap
Subcellular localization info from https://humancellmap.org/ and Go et al. 2021.
* nusiance_genesets
Genesets from HUGO.
* prism
Compound sensitivity data from https://depmap.org/repurposing/ and Corsello et al 2020.

## interim
This folder contains MATLAB files (.mat) that are the output of the dual-graph regularized dictionary learning applied with this codebase (https://github.com/joshbiology/graph_dictionary_learning) on preprocessed input data contained here (10.6084/m9.figshare.14963561).
* depmap_deep
Input: DepMap data. Output: sweep over a few values of k, at t=4 and multiple random seeds.
* depmap_denoise
Input: DepMap data with added noise. Output: Dictionaries (k=220, t=4) learned on noisy DepMap data.
* depmap_grid
Input: DepMap data. Output: sweep over many values of k and t.
* depmap_wide
Input: DepMap data. Output: sweep over many values of k, and t = 4 and the same random seed.
* durocher
Input: genotoxic fitness data (Olivieri et al 2020). Output: sweep over many values of k, at t=1,2.
* durocher_no_graph
Input: genotoxic fitness data (Olivieri et al 2020). Output: sweep over many values of k, at t=1,2, with alpha and beta set to 0.
* synthetic
Input: Synthetic fitness data.

VersionsV1: Initial commitV2: Cached additional intermediate data in the interim folder.