Smith-Waterman Algorithm

ℹ️ Important communication ℹ️

Dear users, I have included this codebase in a new and updated package called seqalign. It aims to become a collection of sequence alignment algorithms. For now it includes Smith-Waterman and Needleman-Wunsch and I plan on expanding it in the future.

---

1. Foreword
2. Getting started
3. SWaligner
4. Defaults
5. Usage
6. Alignment result

Foreword

The Smith-Waterman algorithm is primarily used for local alignment of string sequences from biological datasources (DNA, RNA, protein). I used this algorithm, with good outcomes, for scoring the distance of user inputs from an array of predefined standards.

If you intend to use this code to compare text strings you will need to preprocess your data in order to remove spaces. Punctuation does not compromise the alignment and can be kept.

Getting started

Install the package from npm:

$ npm install --save graphic-smith-waterman

Import the package in your project:

const SWaligner = require('graphic-smith-waterman')

SWaligner

SWaligner is a factory, you can create many aligners with different parameters and re-use each one multiple times. An aligner is configurable with the following parameters (all of them are optional):

• similarityScoreFunction: takes two characters (string) as input and returns a similarity score (integer).
• gapScoreFunction: takes one positive integer as input (gap length) and returns a score (integer).
• gapSymbol: a custom character (string) used to represent gaps in the alignment.
• directions: enum object used to define direction codes for the traceback matrix.

Tip: Higher scores for gaps means higher chances of having one inserted. Generally you should choose a function that gives higher scores to shorter gaps.

Defaults

Here are the default values for the aligner options:

const similarityScoreFunction = (char1, char2) => (char1 === char2 ? 2 : -1);
const gapScoreFunction = k => -k;
const gapSymbol = '-';
const directions = Object.freeze({
    NONE: 0,
    DIAGONAL: 1,
    LEFT: 2,
    UP: 3,
});

Generally, you should not have the need to change the directions enum, but if you need to carry out operations on the traceback matrix yourself, you can define your custom characters, remember:

• It is not necessary to freeze the custom directions object but it is recommended.
• Do not change enum keys (i.e. NONE, DIAGONAL, LEFT, UP) or the algorithm will not work.

Usage

Instantiating SWaligner returns an aligner object which exposes an align method. align accepts the two strings to align as input:

const SWaligner = require('graphic-smith-waterman')

const defaultAligner = SWaligner();
const customAligner = SWaligner({
  gapScoreFunction: x => x / 2,
  gapSymbol: '~',
})

const defaultResult = defaultAligner.align('insertion', 'deletion');
const customResult = customAligner.align('insertion', 'deletion');

console.log(defaultResult.alignment)
// > ertion
// > e-tion

console.log(customResult.alignment)
// > inse~~rtion
// > ~~~ele~tion

Alignment result

The align method returns an object with the following properties:

• score <int>: alignment score.
• originalSequences Array<str>: original input sequences.
• alignedSequences Array<str>: locally aligned sequences.
• scoringMatrix Array<Array<int>>: alignment scores matrix.
• tracebackMatrix Array<Array<int>>: alignment traceback directions matrix.
• coordinateWalk Array<Array<int>>: coordinate walk from the traceback matrix.
• alignment <str>: printable visual alignment string.