I have a file composed of three objects, name y which is a numeric vector of length 2805 (composed only of -1 and 1, where -1 means exon is constitutively spliced, and +1 means alteatively spliced. Then, s1, and s2 are DNAStringSet objects, which both contain 2805 sequences of length 201.
I want to use the kebabs package from Bioconductor, to train a model that predicts alteative splicing. Though, I don't know how can I combine the two sequences, and what are some available keel types.
And afterwards, I have two other data sets with sequences, where I need to use this trained model to predict their label, whether they are +1 or -1.
I can do something like this:
library(kebabs)
spKeel6 <- spectrumKeel(k=6)
promSim <- getKeelMatrix(spKeel6, s1)
But, how can I combine s1 and s2 that I have (I also have the above mentioned y)? How can I train a model with kebabs? And afterwards, how can I predict label for some other sequence, let's say s1t?