Hello guys.
As you can see, data are prepared in the ./dataset
. The method to read the data is also ready. I'm building a toy VAE to learn the origin data. Just for fun.
This part collects any thoughts/plan about the main structure.
We could maintain a doc to keep track of prerequisites. As far as I know, some packages we might need:
scanpy
scipy
geomloss
umap
optuna
numpy
sqlite
pandas
sklearn
torch
Our model use the schibinger dataset. This dataset is embryo sequencing profiles. We can use scipy
to read the dataset. Sample already prepared to read it. For other dataset, some already contained in the scanpy.
I'm building a VAE. also we will test other VAE's performance.
scVAE, ZINVWAVE