Erik Hartman

I’m a PhD student in computational biology at Lund University, currently a visiting research scholar in César de la Fuente’s Machine Biology Group at the University of Pennsylvania.

I’m interested in how mathematics and machine learning can help us understand biology, from interpreting molecular data to designing proteins and peptides. A particular interest is encrypted peptides: fragments of proteins that can have biological functions of their own. I study how these fragments are generated and what roles they may play across the tree of life.

Erik Hartman

Research

I study how proteins give rise to bioactive peptides, and build computational methods to discover and design them.

Encrypted peptides

I investigate bioactive peptides released through protein degradation, particularly antimicrobial peptides and their role in host defense.

A folded protein is cleaved into peptide fragments, with one candidate fragment outlined.
Protein cleavage and candidate peptide discovery. Schematic.

Protein and peptide design

I develop methods for protein and peptide design using machine learning, Bayesian optimization, and biological information.

Candidate sequences are evaluated by a computational model, ranked, and used to guide the next design round.
An iterative approach to sequence design. Schematic.

Protein degradation

I use peptidomics, graph models, and interpretable machine learning to study protein degradation and molecular changes during infection.

A directed graph traces sequential cleavage from longer peptides to shorter fragments.
Representing peptide degradation as a graph. Schematic.

Selected publications

All publications

Software

GitHub

BoPep

Bayesian optimization for exploring protein and peptide sequence space.

Sequence design

Degradation graphs

Model and analyze sequential protein degradation from peptide measurements.

Peptidomics

BINN

Build and interpret neural networks informed by biological pathways.

Interpretable ML

pepnets

Cluster related peptides to reveal patterns in large peptidomics datasets.

Peptidomics

FK-RFdiffusion

Guide protein generation toward structural and biochemical objectives.

Protein design

ProteoCFM

Reconstruct proteomic disease trajectories with conditional flow matching.

Proteomics

Writing

All writing

Frequently asked questions

Answers to questions from family and friends about AI and my research.

The role of protein degradation during infection

A background to my PhD projects

The broken game of academic research: how journals are shifting the incentives of researchers and ruining science

the flawed academic system through the lens of game theory