5 Case studies

We present two distinct use cases of seqme. Extensive details are provided in the Supplementary Text, available as supplementary data at Bioinformatics Advances online.

5.1 Benchmarking antimicrobial peptide design models

In this case study, we evaluate the performance of five existing generative models for antimicrobial peptide (AMP) design. Based on the five metrics employed, we find that there is a trade-off in the generative model’s designs diversity, fidelity to natural peptides, and having antimicrobial properties. Indeed, no single model outperforms all others across the evaluated metrics, as shown by Pareto ranking the models’ calculated metrics using seqme.

5.2 Acquiring high-quality mRNA data

In this case study, the goal is to select a subset of mRNA sequences from a larger pool for subsequent wet-lab evaluation, subject to a fixed sequence budget. This is a common scenario when selecting designed sequences from generative or sequences mined from biological databases. To achieve this, we run a genetic algorithm to select a subset of mRNAs from a larger pool. The fitness of mRNA subsets is evaluated using the metrics FKEA and the stability. By running a genetic algorithm primarily composed of functionality provided by seqme, we obtain more informative training data than we would by selecting sequences at random.