Retrieving contribution of genes to axes

[SMirca]

With PCA, it is possible to get the contribution of the genes to the different axes. Is there a way to get this information with the PCoA of CIDR?

[hojwk]

Unfortunately this is not possible from PCoA (not PCA). The input of PCoA is a pairwise dissimilarity matrix, not a gene expression matrix.

[SMirca]

Is there any way to get the contribution of the genes to the CIDR clustering?

[hojwk]

Sorry, CIDR does not have that functionality. You could consider performing differential expression analysis after you have obtained a suitable clustering in CIDR.

[SMirca]

Thank you for your reply and suggestion. I was wondering also if from CIDR, it's possible to get the new expression table with the imputed values for the dropout candidates and real values for the non-candidates. If I correctly understood how CIDR works, CIDR uses that new expression table to calculate the dissimilarity matrix. I would like to have that new expression table

[hojwk]

The approach implemented in CIDR is called 'implicit imputation'. Imputation is only perform within the calculation of pairwise dissimilarities. We do not (and are unable to) compute a value per drop-out candidate. If you want to perform imputation, there are other published tools to do that. CIDR is designed for ultrafast dimensionality reduction and clustering.

[SMirca]

Thank you for your reply. I do not want to specifically perform an imputation. I want to understand how the imputation step is done by CIDR, what values are imputed within the calculation of pairwise dissimilarities, exactly how CIDR decides which value to impute. I'm not very familiar with codes. So, the CIDR Rscript is not very clear for me about how the imputation is done and how the dissimilarity matrix is calculated.

Thank you for your help

[hojwk]

The details of the implicit imputation steps are explained in this paper: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-017-1188-0

[SMirca]

Thank you