Genome characterization through dichotomic classes: an analysis of the whole chromosome 1 of a. thaliana

In this article we show how dichotomic classes, binary variables naturally derived from a new mathematical model of the genetic code, can be used in order to characterize dierent parts of the genome. In particular, we analyze and compare dierent parts of whole chromosome 1 of Arabidopsis thaliana: genes, exons, introns, coding sequences (CDS), intergenes, untrans- lated regions (UTR) and regulatory sequences. In order to accomplish the task we encode each sequence in the 3 possible reading frames according to the denitions of the dichotomic classes (parity, Rumer and hidden). Then, we perform a statistical analysis on the binary sequences. Interestingly, the results show that coding and non-coding sequences have dierent patterns and proportions of dichotomic classes. This suggests that the frame is important only for coding sequences and that dichotomic classes can be useful to recog- nize them. Moreover, such patterns seem to be more enhanced in CDS than in exons. Also, we derive an independence test in order to assess whether the per- centages observed could be considered as an expression of independent random processes. The results conrm that only genes, exons and CDS seem to possess a dependence structure that distinguishes them from i.i.d sequences. Such in- formational content is independent from the global proportion of nucleotides of a sequence. The present work conrms that the recent mathematical model of the genetic code is a new paradigm for understanding the management and the organization of genetic information and is an innovative tool for investigating informational aspects of error detection/correction mechanisms acting at the level of DNA replication.