Engineering Precision
Pharmacology

At Harmonic Discovery we leverage machine learning to develop next generation medicines with precision pharmacology
Learn More About Our Science
Current
medicine

Current paradigm of drug discovery favors designs that target a single biological mechanism.

Current
medicine

Unexpected off-targets cause side effects.

Current
medicine

Misses additional targets that might be helpful.

Harmonic
Discovery

Removes off targets.

Harmonic
Discovery

At Harmonic Discovery, we are creating a new generation of therapeutics that embrace the complexity of disease

Current drug discovery aims for selectivity

Many drugs are often designed against a single protein target in mind

Yet often fails to address specific anti-targets

Off-target interactions occur, some off target interactions are anti-targets that can drive adverse reactions

And key secondary targets are out of scope

Additional targets can enhance efficacy, mitigate toxicities and shut down resistance pathways

Our platform dials out specific anti-targets

Our generative chemistry platform enables us to identify the correct modifications to molecules that tune out toxic off-targets

And enables
rational multi-targeted drug discovery

By tuning out anti-targets and tuning-in additional targets, we are engineering medicines with precision pharmacology- medicines that can address the complexity of disease

our platform

Built on a bedrock of peer-reviewed science and industry expertise

Our therapeutics are the result of integrating several layers of information, from single point mutations in protein sequence, through 3-dimensional conformational changes of protein structure, to systematic changes in protein gene expression

dive into our Platform
Leveraging multiple data types for improved compound-kinase bioactivity prediction

In this work, we built a model to integrate several bioactivity data types to improve compound-kinase bioactivity prediction

Investigating the Conformational Landscape of AlphaFold2-Predicted Protein Kinase Structures

In this work, we investigated the conformational space and utility of protein kinase structures generated by AlphaFold2

Preference Optimization for Molecular Language Models

In this work, we show that we can use Direct Preference Optimization to align generative chemistry models with medicinal chemist preferences

Built by aN interdisciplinary team

Our team represents the intersection of experts across medicinal chemistry, cheminformatics, data science and machine learning

Join us ↗

We look for passionate innovators, radical
thinkers and collaborative builders

Backed by the best

We are supported by several of the best investors operating at the interface of technology, biology and medicine