Formulation and buffer screening
Formulation at discovery scale is not the full development exercise. It is a smaller question: which buffer, pH, and additives keep this molecule monomeric, potent and pipettable long enough to finish the program, and which candidate in the panel is going to be difficult later.
The variables that move the result
pH is the biggest lever. Colloidal stability is generally worst near the isoelectric point, where net charge is low and molecules associate. Screening a range on either side of the target's pI, in steps, usually finds a clear optimum.
Buffer species matters beyond pH. Histidine, acetate, citrate and phosphate differ in how they interact with proteins, and phosphate in particular behaves differently on freezing, where its pH shifts as one component crystallizes.
Ionic strength screens charge interactions. Low salt favors some molecules and destabilizes others; it is not predictable enough to skip testing.
Excipients. A nonionic surfactant protects against interfacial stress from agitation and pumping. Sugars and polyols stabilize during freezing and drying. Amino acids such as arginine can suppress aggregation and reduce viscosity at high concentration.
A screen that fits discovery
Take the shortlist, put each candidate into a matrix of conditions at a fixed concentration, and measure three things: thermal unfolding and aggregation onset, monomer content by size exclusion after a defined stress, and visible or sub-visible particle formation.
For stress, choose what the molecule will actually experience: freeze-thaw cycles, agitation, and a short accelerated hold at elevated temperature. An accelerated hold is a screening tool rather than a prediction of shelf life, because the degradation routes at elevated temperature are not always the ones that dominate at storage temperature.
Run the panel in one experiment so conditions are comparable, and include a reference molecule.
Concentration changes everything
Aggregation, viscosity and self-association are concentration-dependent, and the concentration that matters is the final one, not the convenient one. A molecule that is well behaved at one milligram per milliliter can be unusable at a hundred. If the program will need high concentration, screen there, or at least measure a self-interaction parameter that indicates how the molecule will behave when crowded.
What to hand over
Record the chosen buffer with every aliquot, and use it consistently across the panel. Most of the value at this stage is comparability: if every candidate sits in its own buffer, the stability ranking is a ranking of buffers, and the conclusion about which molecule to advance is wrong in a way nobody can see.
The other output is a list. Candidates that needed unusual conditions to behave, that aggregated on freeze-thaw, or that showed particles after agitation are flagged for development, because those are the molecules that turn into problems after a program has committed to them.