Feature Request- Conditional GPLVM

[willtownes]

In the Nature Methods paper it suggests it would be possible to include covariates in the scLVM procedure but I don't think it is fully implemented yet in the gp_clvm function. Just wanted to say I'm very interested in this application and look forward to seeing it as a new feature.

[flophys]

Hi.

Yes, this is possible and fully implemented. For R (recommended interface) the functionality is illustrated in the tutorial. In the python version the procedure is very similar for fitting the conditional GPLVM with the gp_clvm function. If you just have covariates which you would like to include for the variance decomposition and/or correlation analysis as described in the paper you can just pass them to the respective methods. In this case K is a list of all covariance matrices, corresponding to inferred factors and as well as known covariates. If you'd like to include batch, for example, you can construct the corresponding covariance matrix using a delta kernel.
I will include this in the python vignette soon, too.

Florian

[AmmarMalik93]

I will include this in the python vignette soon, too.

Hi Florian,
I am unable to find this functionality in the python vignette. (https://github.com/PMBio/scLVM/blob/master/tutorials/tcell_demo.ipynb)

Considering this post is about 3-4 years old, so I am guessing it is already there but for some reason I am not able to find it. So, if you would be kind enough to point me in the right direction.

Forgive my naivety as I am a beginner in bioinformatics/computational biology field.

Ammar