Revealing the Presence of a Symbolic Sequence Representing Multiple Nucleotides Based on K-Means Clustering of Oligonucleotides.

Abstract

In biological systems, a few sequence differences diversify the hybridization profile of nucleotides and enable the quantitative control of cellular metabolism in a cooperative manner. In this respect, the information required for a better understanding may not be in each nucleotide sequence, but representative information contained among them. Existing methodologies for nucleotide sequence design have been optimized to track the function of the genetic molecule and predict interaction with others. However, there has been no attempt to extract new sequence information to represent their inheritance function. Here, we tried to conceptually reveal the presence of a representative sequence from groups of nucleotides. The combined application of the K-means clustering algorithm and the social network analysis theorem enabled the effective calculation of the representative sequence. First, a “common sequence” is made that has the highest hybridization property to analog sequences. Next, the sequence complementary to the common sequence is designated as a ‘representative sequence’. Based on this, we obtained a representative sequence from multiple analog sequences that are 8–10-bases long. Their hybridization was empirically tested, which confirmed that the common sequence had the highest hybridization tendency, and the representative sequence better alignment with the analogs compared to a mere complementary.