Get the sparse inverse covariance matrices per tumor type.
Perform PCA/NMF in the space of sparse covariance matrices, e.g TangentPCA
"https://github.com/geomstats/geomstats"
Generate 2D embeddings of archetypical covariances "graphs", and plot differences of specific tumor type covariance matrices to these archetypes.
Problem: how to anchor e.g. forcee directed 2D embeddings as e.g. in tsne? Try with warm start from archetype solution.
Sparse inverse covariance estimation with the graphical lasso - Jerome Friedman, Trevor Hastie, Robert Tibshirani Biostatistics, Volume 9, Issue 3, July 2008, https://academic.oup.com/biostatistics/article/9/3/432/224260 some code - https://scikit-learn.org/stable/auto_examples/applications/plot_stock_market.html#sphx-glr-auto-examples-applications-plot-stock-market-py
Get the sparse inverse covariance matrices per tumor type. Perform PCA/NMF in the space of sparse covariance matrices, e.g TangentPCA "https://github.com/geomstats/geomstats"
Generate 2D embeddings of archetypical covariances "graphs", and plot differences of specific tumor type covariance matrices to these archetypes.
Problem: how to anchor e.g. forcee directed 2D embeddings as e.g. in tsne? Try with warm start from archetype solution.