E. Granjeon et P. Tarroux, DETECTION OF COMPOSITIONAL CONSTRAINTS IN NUCLEIC-ACID SEQUENCES USING NEURAL NETWORKS, Computer applications in the biosciences, 11(1), 1995, pp. 29-37
We describe in this paper a neural network method for the detection of
compositional constraints in introns and exons. The first part of the
algorithm (learning phase) consisted in presenting examples of intron
and exon sequences to the network and in modifying its connections us
ing the backpropagation algorithm. Previous connectionist methods achi
eved the learning of exons and introns using the latter as negative ex
amples to the former. However, we chose to learn introns and exons joi
ntly, using junk DNA as a common counter-example. In a second part (ge
neralization phase), we rested the neural networks in the search for e
xons and introns in the human globin cluster. Their performances were
also checked on the classification of unknown examples. As with the pr
evious approaches, this technique discriminates introns and exons: val
ues of the correlation coefficients are respectively 0.50 and 0.64 for
the best achieved network. Moreover, using junk DNA sequences in the
learning phase allows one to detect constrained regions inside the int
ron and the exon sequences (i.e. sequences that differ, by their nucle
ic acid compositions, from junk DNA). The application of our approach
could be useful in the study of the internal organization of these seq
uences.