Skip to content

Fitness Functions

This section details the objective functions available for the genetic algorithm (GA). Each fitness module evaluates a specific biological or biophysical property of the query sequence, returning a numerical score that drives the optimization landscape.

Overview

The optimization engine accepts a list of fitness objects. During the evolutionary process, these metrics are computed to rank candidate sequences.

Common Parameters: Every fitness function accepts the following optional parameter:

Parameter Type Default Description
weight float 1.0 Determines the magnitude and direction of the fitness vector. Positive values maximize the metric; negative values minimize it.

Fitness list

1. RNA Pseudoyield

Description A metric based on deep learning predicting the yield of in vitro RNA transcription (IVT). It calculates a measure of IVT yield of full-length RNA molecules.

Arguments * weight (float, optional): Optimization weight.


2. Manufacturability

Description A metric that takes into account rules for RNA manufacturability such as GC content and codon usage.

Arguments * weight (float, optional): Optimization weight.


3. CAI (Codon Adaptation Index)

Description Quantifies the conformity of the sequence's codon usage to the synonymous codon usage bias of a reference set of highly expressed genes. High CAI values are theoretically linked to efficient translation elongation and reduced metabolic burden.

Arguments * weight (float, optional): Optimization weight.


4. U% (Uridine Content)

Description Calculates the percentage of uridine nucleotides within the sequence. In the context of mRNA therapeutics, reducing U-content (U-depletion) is a strategy to minimize immunogenicity via Toll-like receptor (TLR) evasion and to improve mRNA stability.

Arguments * weight (float, optional): Optimization weight.


Technical Summary

ID Name Range
ppy RNA Pseudoyield Unbounded
rna_manufact RNA Manufacturability (0, 1]
cai Codon Adaptation Index (0, 1]
u_perc Uridine Percentage (0, 1]