A Fast Template Based Heuristic For Global Multiple Sequence Alignment

TL;DR

This paper introduces a fast, template-based heuristic for global multiple sequence alignment that incorporates feature classification and weighting, improving biological accuracy by leveraging structural, functional, and evolutionary data.

Contribution

It provides a novel mechanism for explicitly specifying feature types and weights, and develops a scoring model based on segment conservation for more accurate alignments.

Findings

Enables explicit feature classification and weighting in template-based heuristics.

Defines a scoring model based on segment conservation rather than single residues.

Proposes a fast progressive alignment heuristic for efficient global MSA construction.

Abstract

Advances in bio-technology have made available massive amounts of functional, structural and genomic data for many biological sequences. This increased availability of heterogeneous biological data has resulted in biological applications where a multiple sequence alignment (msa) is required for aligning similar features, where a feature is described in structural, functional or evolutionary terms. In these applications, for a given set of sequences, depending on the feature of interest the optimal msa is likely to be different, and sequence similarity can only be used as a rough initial estimate on the accuracy of an msa. This has motivated the growth in template based heuristics that supplement the sequence information with evolutionary, structural and functional data and exploit feature similarity instead of sequence similarity to construct multiple sequence alignments that are biologically more accurate. However, current frameworks for designing template based heuristics do not allow the user to explicitly specify information that can help to classify features into types and associate weights signifying the relative importance of a feature with respect to other features. In this paper, we first provide a mechanism where as a part of the template information the user can explicitly specify for each feature, its type, and weight. The type is to classify the features into different categories based on their characteristics and the weight signifies the relative importance of a feature with respect to other features in that sequence. Second, we exploit the above information to define scoring models for pair-wise sequence alignment that assume segment conservation as opposed to single character (residue) conservation. Finally, we present a fast progressive alignment based heuristic framework that helps in constructing a global msa efficiently.