Preface

DNA is simultaneously a molecule, a durable information medium, and the substrate of a distributed chemical control system. Those roles make it a remarkable subject for computer scientists, but they also create a common source of confusion: biological information processing, wet-lab DNA computation, and machine learning on genomic sequences are related fields, not interchangeable names for the same technology.

This book builds a careful bridge among them. It begins with nucleotides and the central dogma of molecular biology, develops strand displacement and chemical reaction networks as algorithms, studies gene regulation as conditional computation, and then supplies the machine-learning background needed to understand modern genomic foundation models. The aim is not to claim that biology has already solved artificial intelligence. The aim is to give students enough biology, computation, and mathematics to ask better questions and design reproducible experiments.

Audience. The text is written for undergraduate students in computer science, biology, engineering, and data science. Introductory programming, calculus, linear algebra, and probability are helpful. No prior molecular biology is assumed.

Evidence discipline. Established biological mechanisms are identified as biological facts and cited to the literature. Results from other laboratories are external results. Proposed computational mechanisms remain hypotheses until an implementation, a frozen evaluation, and a reproducible artifact support a stronger label.

Relationship to the companion volume. The companion research monograph, DOGMA: DNA-Organized Genomic Model Architecture, explores a specific non-transformer architecture inspired by persistent state, regulation, local interaction, expression, and population selection. Readers can study this foundations volume independently.

Wenyan Qin
July 2026