Training and evaluating a variational autoencoder for pan-cancer gene expression data
The release tracks recent developments in evaluating tybalt models. The release includes an evaluation of two independent 2 hidden layer models (one with 100 and the other with 300 hidden layer size) given in this notebook as well as a latent space arithmetic analysis in high grade serous ovarian cancer subtypes in this notebook.
In each analysis, the model was trained using TCGA pan-cancer gene expression data.
This tag is associated with analyses submitted to the 2017 NIPS conference.
The release incorporates an end-to-end training and evaluating of a variational autoencoder applied to model TCGA pan-cancer gene expression data. Future releases will stabilize the pipeline and also introduce novel modeling features and evaluation techniques.
Minor Updates:
0.1.2 - Upgrade to HGSC interpolation analysis
0.1.2 - Give the model the name Tybalt
The release incorporates an end-to-end training and evaluating of a variational autoencoder applied to model TCGA pan-cancer gene expression data. Future releases will stabilize the pipeline and also introduce novel modeling features and evaluation techniques.
Minor Updates: 0.1.1 - Fix legibility of parameter sweep figures
The release incorporates an end-to-end training and evaluating of a variational autoencoder applied to model TCGA pan-cancer gene expression data. Future releases will stabilize the pipeline and also introduce novel modeling features and evaluation techniques.