Convolving DNA using two-dimensional Hilbert curve representations


Nov 03, 2017 (modified: Dec 12, 2017) ICLR 2018 Conference Blind Submission readers: everyone Show Bibtex
  • Abstract: The folding structure of the DNA molecule combined with helper molecules, also referred to as the chromatin, is highly relevant for the functional properties of DNA. The chromatin structure is largely determined by the underlying primary DNA sequence, though the interaction is not yet fully understood. In this paper we develop a convolutional neural network that predicts key determinants of chromatin structure from primary DNA sequence. Key in our approach is the transformation of the input data from a sequence to an image using the Hilbert curve, which has several desirable properties from a biological point of view. Our experiments show that the method outperforms several existing methods both in terms of prediction accuracy and training time.
  • TL;DR: A method to transform DNA sequences into 2D images using space-filling Hilbert Curves to enhance the strengths of CNNs
  • Keywords: DNA sequences, Hilbert curves, Convolutional neural networks, chromatin structure