From Eqtnminer

Jump to: navigation, search



eqmr-expca performs principal component analysis onto a given matrix using either standard full single value decomposition or using an EM algorithm that can better handle missing data and that can also reduce significantly the computational burden for large matrices when only the few first axes are needed.

Besides the program can also perform a permutation procedure in order to generate the empirical distribution of the eigenvalues under the null hypothesis that there is no significant covariance structure due to the presence of hidden factors.

Presently, only expression matrices outputted by eqmr-fexp can be used. Future development will allow to start from a custom matrix provided via a standard text file.


Short Long Description
-e --expmat The expression matrix file as generated by eqmr-fexp (binary format)
-o --output The output file
-k --kmax The maximum number of axes for the PCA (default is min(#row, #col))
-p --permut The number of permutations to do to compute PC p-values
--scale Scale the features by their variance
--sample Use samples as features


See Also

warning.png"{{{mainsourcefile}}}" cannot be used as a page name in this wiki.

Facts about Eqmr-expcaRDF feed
Main source filewarning.png"{{{mainsourcefile}}}" cannot be used as a page name in this wiki.
Program categoryData Analysis  +
Program languageC  +
Program nameeqmr-expca  +
Program titlePCA on the expression dataset
Personal tools