Memory Matching Networks for Genomic Sequence Classification

TL;DR

This paper introduces Memory Matching Networks (MMN) that utilize a learned memory bank of motifs to classify DNA sequences as protein binding sites, addressing the complexity of motif identification.

Contribution

The work presents a novel MMN approach that dynamically encodes motifs and matches sequences for classification, improving over manual motif construction methods.

Findings

Effective classification of DNA sequences as binding or nonbinding sites.

Memory bank of motifs enables dynamic and accurate reasoning.

Outperforms traditional motif-based methods.

Abstract

When analyzing the genome, researchers have discovered that proteins bind to DNA based on certain patterns of the DNA sequence known as "motifs". However, it is difficult to manually construct motifs due to their complexity. Recently, externally learned memory models have proven to be effective methods for reasoning over inputs and supporting sets. In this work, we present memory matching networks (MMN) for classifying DNA sequences as protein binding sites. Our model learns a memory bank of encoded motifs, which are dynamic memory modules, and then matches a new test sequence to each of the motifs to classify the sequence as a binding or nonbinding site.