Map-Reduce Parallelization of Motif Discovery

TL;DR

This paper proposes a Map-Reduce based parallelization approach for motif discovery in DNA sequences using HOMER, aiming to improve scalability but facing challenges with motif quality.

Contribution

It introduces a novel parallelization methodology for HOMER using sub-sampling and Map-Reduce, enabling potential scalability improvements in motif discovery.

Findings

Marginal speed gains achieved with parallelization.

Significant quality loss in discovered motifs.

Method demonstrates potential for scalable motif discovery.

Abstract

Motif discovery is one of the most challenging problems in bioinformatics today. DNA sequence motifs are becoming increasingly important in analysis of gene regulation. Motifs are short, recurring patterns in DNA that have a biological function. For example, they indicate binding sites for Transcription Factors (TFs) and nucleases. There are a number of Motif Discovery algorithms that run sequentially. The sequential nature stops these algorithms from being parallelized. HOMER is one such Motif discovery tool, that we have decided to use to overcome this limitation. To overcome this limitation, we propose a new methodology for Motif Discovery, using HOMER, that parallelizes the task. Parallelized version can potentially yield better scalability and performance. To achieve this, we have decided to use sub-sampling and the Map Reduce model. At each Map node, a sub-sampled version of the input DNA sequences is used as input to HOMER. Subsampling at each map node is performed with different parameters to ensure that no two HOMER instances receive identical inputs. The output of the map phase and the input of the reduce phase is a list of Motifs discovered using the sub-sampled sequences. The reduce phase calculates the mode, most frequent Motifs, and outputs them as the final discovered Motifs. We found marginal speed gains with this model of execution and substantial amount of quality loss in discovered Motifs.