Ometer — File transformations

Transposition

Sometimes it is useful to be able to transpose a file, such that the meaning of variables and samples is reversed. Ometer provides an option that allows this operation to be carried out easily:  -m=transpose or --method=transpose

The result of this operation will always be a file in the preferred format of ometer, variables on rows and samples on columns (-f=1), independently of the format of the input file.

Ometer will always remove the classification of samples when it is carrying out a transposition. This is because the transposition will change the meaning of samples to variables, and variables are never classified (by definition what one classifies are the samples). In fact, this property is why this option exists, otherwise it would be enough to change the file format option  -f to consider the transpose.

Transposition is especially useful when one is interested in finding the coordinates of the variables on the principal component space in PCA analysis (sometimes known as the loadings for the variables, rather than the more usual loadings for the PCs). To achieve this one transposes the file, then carries out PCA on the resulting file; the coordinates are then the projected data (variables projected onto the PCs), which can be found on the html report file and on the  *-pca-proj.dat file.

Removal of missing data

This feature is not yet available, it may be implemented in a future version of ometer (if one is ever made). Several methods may be used to fill in the missing data values, or to remove variables or samples with missing values.

back to Ometer home page