I'm trying to analyze gene expression (60 genes) harvested form 150 cells using single cell PCR.
I'd like to run a pca on these data but I have around 20 % of zero values (from 60 to 0 % depending on genes and cell) that could be true zero values (no expression or very low expression since we may have sub-groups in our sampling) or some of them could be missing values. PCA doesn't like missing values. How can I handle this ?
Leaving these values at zero ? Doing PCA on the correlation matrix (if possible in R) without the zero values ? Pca on proportion of zero values ?
Thanks
More Fabien Tell's questions See All
Similar questions and discussions