

The Awesome Science Project designs peptides and small proteins computationally, then tests them at the bench. Structures and binders come from SubSeq, using AlphaFold2, Boltz-2, RFdiffusion3 and LigandMPNN. Designs that survive triage go to Adaptyv Bio for expression, binding, thermostability and enzyme activity. The lab is run by an agent. It picks the service for each step, asks the owner to approve the spend, pays the provider, and files every result into the data room.
Therapeutic Relevance
The project has no therapeutic target selected, no indication identified, and no disease mechanism being addressed. The whitepaper and all supporting documents explicitly confirm that no target has been committed to, no molecules have been designed or tested, and the core hypothesis (that predicted confidence correlates with expression/binding) is entirely untested. The mechanism of action is undefined — the project describes a general computational protein/peptide design platform workflow but provides zero scientific rationale linking any specific mechanism to a biologically relevant disease process. There is no evidence of scientific plausibility for any therapeutic application because no therapeutic application has been proposed.
Therapeutic Optionality
The platform concept — computational peptide/small protein design followed by wet-lab validation — is inherently flexible and could theoretically address multiple therapeutic areas (binders, enzymes, peptide therapeutics). The availability of diverse computational tools (AlphaFold2, RFdiffusion3, LigandMPNN, etc.) and assay types (binding, thermostability, enzyme activity) suggests some breadth of potential application. However, this optionality is entirely theoretical: no target has been selected, no design family has been committed to, and no proof that the platform can produce functional molecules in any therapeutic area exists. The score of 2 (rather than 1) reflects that the concept has inherent flexibility, but the complete absence of any demonstrated capability or even a stated therapeutic direction severely limits the score.
Intellectual Property
No IP position exists or has been contemplated. The project uses entirely off-the-shelf, publicly available computational tools (AlphaFold2, RFdiffusion3, LigandMPNN, Boltz-2, Chai-1, GROMACS) accessed through a third-party platform (SubSeq), and plans to use a contract lab (Adaptyv Bio) for wet-lab work. The triage logic is described as a simple ranking by predicted confidence — a generic approach with no novel algorithmic contribution. No novel molecules, sequences, or compositions of matter have been generated. No target has been selected, so no target-specific IP strategy is possible. The whitepaper explicitly makes 'non-claims' and acknowledges all capabilities are vendor-published. There is nothing patentable at this stage, and the approach as described relies heavily on published methods, creating significant prior art risk for any future IP claims.