What is Protein Folding?

Proteins are molecular machines that perform nearly every function in your body. They're built from chains of amino acids, but a chain alone is useless—it must fold into a precise three-dimensional shape to work. This shape determines everything: whether an enzyme can catalyze a reaction, whether an antibody can recognize a virus, whether a signal can pass between neurons.

When proteins misfold, the consequences can be devastating. Misfolded proteins can aggregate into toxic clumps, lose their function entirely, or gain harmful new properties. Understanding how proteins fold—and how mutations cause them to misfold—is one of the most important challenges in modern biology.

For decades, determining a protein's structure required years of laboratory work. In 2020, DeepMind's AlphaFold changed everything. This AI system can predict protein structures from amino acid sequences alone, with accuracy rivaling experimental methods. What once took years now takes minutes.

The Diseases We Study

Clarity Protocol focuses on neurodegenerative diseases—conditions where protein misfolding destroys the brain over time.

Why This Matters

Neurodegenerative diseases represent one of the greatest healthcare challenges of our time.

By 2050, the number of people with dementia is projected to reach 139 million. Current treatments only manage symptoms—none can stop or reverse the underlying disease. We need to understand these proteins at the atomic level to develop therapies that actually work.

Every mutant protein structure we analyze is a potential clue. Understanding how a single amino acid change affects protein folding could reveal why some mutations cause aggressive early-onset disease while others are benign. This knowledge is essential for developing targeted treatments.

How Clarity Protocol Works

Clarity Protocol is an autonomous research pipeline. It continuously monitors scientific literature for newly discovered protein variants, predicts their three-dimensional structures, analyzes the structural consequences of mutations, and generates research summaries—all without human intervention.

The system combines structure prediction with computational analysis and AI-powered interpretation. When a variant is processed, the pipeline doesn't just produce a structure—it compares that structure against wild-type proteins, identifies destabilized regions, and connects findings to relevant published research.

Everything is made publicly available as it's generated. Structures can be explored interactively. Analysis results are presented alongside AI summaries that explain findings in accessible language while maintaining scientific accuracy.

From Structure to Drug Candidate

A protein structure alone is just a starting point. The harder question—the one the pharmaceutical industry spends billions on—is what binds to that structure and changes its behavior. Clarity Protocol extends the folding pipeline with a de novo peptide design stage that proposes short, brand-new amino-acid chains engineered to stick to the sticky parts of disease proteins and block them from aggregating.

Each candidate peptide is generated by a diffusion model (BoltzGen), scored independently by a second model (Boltz-2), and then run through a battery of filters that ask: does it actually bind, does it make real chemical contacts with the target, and would a wet-lab waste money synthesizing it? Only candidates that pass every filter are tagged wet-lab ready. Everything is logged. Sequences are withheld pending IP review.

Pharmacokinetics & Blood-Brain Barrier

A peptide that binds perfectly in a test tube is useless if your bloodstream chews it up in two minutes, or if it can't reach a target that lives inside the brain. The pipeline's final two filters address exactly this:

• Pharmacokinetics (t½, renal, immunogenicity): For every candidate, we estimate its serum half-life against a panel of thirteen circulating proteases, check whether it will be filtered by the kidneys, and flag hydrophobic regions that might trigger an immune response. Failing candidates don't get dropped—instead, the pipeline suggests chemistry modifications (N-terminal capping, lipidation, cyclization, D-amino-acid substitution, PEGylation) that would realistically extend half-life to clinically useful ranges.
• Blood-brain barrier (BBB) strategy: For targets that live in the central nervous system (tau, alpha-synuclein, TDP-43, etc.), a bare peptide almost never crosses the BBB. When the pipeline detects a CNS target, it auto-recommends conjugation to a clinically-relevant shuttle peptide (Angiopep-2, transferrin-receptor binders, ApoE mimetics, glutathione or glucose conjugates) chosen from public literature. For small, charged peptides it also flags intranasal nose-to-brain delivery as an alternative route that bypasses the barrier entirely.

Both gates are advisory by default—they annotate each candidate with a plain-language rationale and suggested chemistry, but never silently reject a design. The point is to give a researcher looking at a candidate card everything they need to decide: is this worth ordering, and if so, what modifications should I ask the synthesis company for?