1
$\begingroup$

I have a bed file which contains DNA sequences information as follow:

**

track name="194" description="194 methylation (sites)" color=0,60,120 useScore=1
chr1    15864   15866   FALSE   894 +
chr1    534241  534243  FALSE   921 -
chr1    710096  710098  FALSE   729 +
chr1    714176  714178  FALSE   12  -
chr1    720864  720866  FALSE   988 -

**

I loaded the bed file in R and named the matrix DataSet. I used the follow code to get the sequences:

mydataSet_Test1<-dataSet[,1:3]

library(BSgenome.Hsapiens.UCSC.hg19)
genome <- BSgenome.Hsapiens.UCSC.hg19
chr<-as.matrix(as.character(mydataSet_Test1[,1]))

#50
start<-as.matrix(as.integer(as.character(mydataSet_Test1[,2]))-50)
end<-as.matrix(as.integer(as.character(mydataSet_Test1[,3]))+50)
Seqs50_Test1<-getSeq(genome,chr,start=start,end=end)

I'm having DNA sequences and for each sequence, I have its Methylation Number (i.e. 800). I have split the data to hypomethylated (methylation < 200, class 0) and hypermethylated (methylation > 800, class 1) and I want to build a model that can predict if a sequence (or a group of sequences) is hypo or hyper - methylated.

I now have a matrix of 358.367 data and 2 columns. The first column is the DNA sequences and the second column is the class for each sequence (0 or 1).

I want to build a classification model in R, using SVM algorithm and 83 features (dinucleotides, trinucleotides, etc.).

Before calculating the features and building the model, are there any steps I should follow for my DNA sequence data (each row of the matrix is a sequence from the human genome), like filtering CpG islands or sequences, homolog reduction or overlapping sequences etc?

As you understand I have a lack in DNA analysis.

$\endgroup$
  • 2
    $\begingroup$ Hi Giannis, could you please describe the classification problem you are trying to solve? $\endgroup$ – Daniel Standage Jun 3 at 14:11
  • $\begingroup$ I'm having DNA sequences and for each sequence, I have its Methylation Number (i.e. 800). I have split the data to hypomethylated (methylation < 200, class 0) and hypermethylated (methylation > 800, class 1) and I want to build a model that can predict if a sequence (or a group of sequences) is hypo or hyper - methylated $\endgroup$ – Giannis Lazaridis Jun 4 at 5:43
  • 2
    $\begingroup$ @GiannisLazaridis Please edit the question to include the last comment, and include the code or guides you have tried $\endgroup$ – llrs Jun 4 at 7:50
  • $\begingroup$ I edited the question $\endgroup$ – Giannis Lazaridis Jun 4 at 11:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.