Machine learning trims dna sequences to keep function while shortening
From DNA Design to DNA Slimming: Auditable Agentic Discovery of a Deletion-Only Designer
Machine LearningNeural and Evolutionary Computing
Summary
DNA sequences used in experiments can be made shorter to save space and effort while keeping their important functions. Most methods change DNA by swapping bases but do not try to remove parts without losing function. The authors define a new task to find a shorter version of a DNA sequence by deleting bases while keeping it working. They built a system called ERA that improved on earlier designs to find these shorter sequences and tested it on several DNA targets with good results.
What this means in practice
- •For synthetic biology developers: Create shorter DNA regulatory sequences that retain function, saving costs and allowing more compact vector design for gene engineering.
- •For bioinformatics engineers: Use an auditable deletion-only designer to generate compact DNA elements by selecting subsequences and verifying compliance with length and order constraints.
Authors
Joel Shor
Abstract
Compact regulatory DNA can free up space in vector payloads, reduce synthesis and assay burden, and expose which sequence features drive predicted activity. Yet most model-based nucleic-acid designers optimize fixed-length sequences through substitutions; they do not ask which bases of an existing functional element can be removed while retaining predicted activity. We define the task of sequence slimming as selecting an exact-length, order-preserving subsequence while retaining activity. Modeled on the design benchmark NucleoBench, we propose a quantitative evaluation for slimming that balances sequence reduction with maintaining function. Each slimmer must return both the subsequence and its source indices, which can be used to verify that the slimmer obeyed task requirements. To our knowledge, this is the first dedicated benchmark of this deletion-only problem. The coding agent Empirical Research Assistant (ERA) then searched over executable designer programs. ERA received the task prompt and a successful substitution-only designer GrAdaBeam as a starting program, and it modified the designer to produce GRADASLIM. We report held-out evaluations for five transcription-factor binding targets, comparing random, greedy, and ERA-guided slimming at 400 and 100 bp. ERA has the highest mean in 9/10 settings. Paired bootstrap intervals for ERA minus greedy are above zero in all five 400-bp settings, below zero in one 100-bp setting, and overlap zero in the remaining four.