A drug discovery pipeline funnels around 5,000 to 10,000 compounds down to ~250, down to ~5, down to one approved drug. The funnel is unforgiving by design. What it should not have, but in practice still does, is a steady leak of compounds that drop out of the funnel for reasons that were knowable much earlier.
Solubility and aggregation are common culprits in that leak. Compounds that look clean at a hit-finding stage, but that are quietly poorly soluble or quietly aggregation-prone, will fail in pre-clinical development or generate misleading SAR that costs the chemistry team months of optimisation. The cost of catching a developability risk at hit-to-lead is small. The cost of catching it after lead optimisation is large. The cost of catching it after pre-clinical work has begun is enormous.
The reason these costs persist is not that discovery teams do not understand the problem. They do. The reason is that the measurement workflows that could catch these risks at hit-finding pace simply did not exist for the molecule classes pharma now actively pursues.
Essentially, we kept seeig hard-won leads get killed by late-arriving physics — the solubility and aggregation data nobody could afford to measure early. That’s why we built ORYL F1. We just demoed it recently at top-five big pharma site: three workflows on one instrument — 10 mM compound QC in compound management, aggregation triage ahead of high-throughput screening, and pre-formulation profiling. One plate-based platform, one method, three points in the discovery-to-pre-clinical chain. Triage-level data and not after the fact, providing teams a higher chance to validate their activity/potency data and de-risk solubility and aggregation early.
Starting with compound stock QC
Compound management is where the data trail starts. Most discovery libraries are stored as nominal 10 mM DMSO stocks, and the word “nominal” is doing a lot of work. Compounds precipitate, aggregate, or partially dissolve during storage and handling, and the stock that gets pulled into a screen is not always the stock that was logged on dispense. If the starting point is wrong, every measurement downstream is wrong with it. Application Note (AN-1002) – DMSO QC covers this case in detail: a non-destructive Go/No-Go QC workflow on ORYL F1, directly in ECHO LDV plates, ~15 minutes per 384 wells, using complementary SHS and LLS readouts to flag problematic stocks before they reach downstream assays.
Cleaning your HTS hit list
Once stocks are clean, the next bottleneck is what gets fed into biophysical assays. Aggregators are expensive – and they are not wanted. They block fluidics, dirty SPR chips, and burn through surface chemistry. They generate false positives in SPR readouts and seed SAR your chemistry team will spend months optimising against — chasing a target that isn’t really there. By the time the issue surfaces in pre-formulation, the chemistry is already in flight. Removing aggregators from the hit list before SPR or other biophysics has been the right thing to do for years; it has also been impractical at hit-list scale, because legacy methods couldn’t match screening throughput. ORYL F1 changes that math: 384 wells in ~15 minutes, ~100× less compound than HPLC-based methods. Aggregation triage becomes a same-day operation, ahead of HTS — not a forensic exercise after.
Dense formulation mapping
By the time a lead reaches pre-formulation, the question shifts. It is no longer “does this work?” — it is “how does this molecule actually behave?” Solubility windows. Aggregation onset. Behaviour at dosing-relevant concentrations. This is where dense-formulation work happens, and where formulation scientists need to understand the molecule well enough to defend choices about salts, polymorphs, excipients, and dosing format. ORYL F1 supports that workflow on the same instrument that ran the compound-stock QC and the aggregation triage upstream — same plate format, same scattering readouts, same data layer. Formulation teams inherit a coherent data trail rather than a patchwork of method-specific results from different instruments at different stages.
The math has broken — and AI is making it urgent
Legacy solubility workflows assume time and compound. Milligram-scale quantities. Method development per compound class. Days to weeks for a clean profile. None of that fits the pace or compound economics of current discovery work — and AI is making the gap wider. Hit lists are larger, libraries more diverse, cycle times shorter. AI-driven prioritisation and design are expected to compress discovery and development timelines by double-digit percentages in the coming years. The constraint is no longer ideas; it is reliable, large-scale, well-measured data to feed those models. Solubility and aggregation are exactly the kind of physical properties they need on hand, at scale, with consistent measurement provenance. That changes what an instrument has to do. Throughput stops being a nice-to-have and becomes the entry condition — thousands of compounds per day is the new floor, not the ceiling. ORYL F1 was built for that. 384 wells in ~15 minutes. ~100× less compound than HPLC-based methods. Two complementary scattering readouts — Second Harmonic Scattering (SHS) and Linear Light Scattering (LLS) — captured in the same measurement, powered by Ultrafast Light Scattering. Built for screening pace — triage throughput, triage compound economics. Discovery and pre-formulation share the same data at triage, not after. The shortlist gets better. The handoff gets cleaner. The pipeline moves.