Dear all,

I´d like to use the MLSeq package (http://bioconductor.org/packages/devel/bioc/html/MLSeq.html) which requires a count table that contains the number of reads mapped to each transcript for each sample and class label information of samples in an S4 class DESeqDataSet format. Of course, this is easy to do if you analyze gene expression data. In my case however I´d like to analyze magnetic resonance imaging (MRI) data (about 10.000 imaging parameters per patient), thus I cannot provide a "count table" (I can only provide the raw values or z-score normalized values of each imaging parameter; usually I build an ExpressionSet with all of my imaging data & phenotypes and this worked well so far). I´d be curious if there is a possibility that MLSeq also works with this kind of input?

Thanks for your help!

I've never used MLSeq before, but if it's "specialized" for rnaseq data, why not just use caret, or one of the ML algorithms that these packages wrap directly?

Thank you for suggesting caret. This also works nice for my purposes!