Ref E3
Computational Service
Confidence Reported
Scoped After Feasibility Review
Service Overview
Alchemical free energy calculation is the most accurate binding prediction routinely available, with reported accuracy approaching one kilocalorie per mole for well-behaved systems. It is also expensive, so it belongs at the point in lead optimisation where a wrong synthesis decision costs more than the calculation. We run relative and, where appropriate, absolute binding free energy calculations with careful perturbation mapping, convergence checking and validation against your existing measured data.
Key Features & Advantages
Validated Against Your Own Series
Before predicting untested compounds, we reproduce measured affinities for compounds you have already made, so the expected error on your system is known rather than quoted from a paper.
Perturbation Maps Designed Carefully
Network topology, cycle closure and intermediate states are planned to give redundancy, which both improves accuracy and exposes individual calculations that have failed.
Convergence Checked Per Edge
Every transformation is assessed for phase-space overlap, hysteresis and convergence; edges that fail are rerun or reported as unreliable rather than averaged into the result.
Scoped To Decisions
We advise where FEP is worth its cost and where MM/GBSA suffices, because applying it indiscriminately wastes budget without improving decisions.
Where This Service Is Used
1Choosing which analogues to synthesise in a lead optimisation cycle
2Predicting the potency effect of a specific substituent change before making it
3Assessing selectivity by calculating binding to the target and to a close off-target
4Quantifying the effect of a resistance mutation on binding affinity
5Prioritising a final candidate shortlist where potency differences are small but consequential
Methods and Techniques Applied
1System preparation with consistent protonation, tautomer and binding mode assignment across the series
2Perturbation network design with cycle closure redundancy and controlled transformation size
3Lambda window and soft-core scheduling, with enhanced sampling for slow degrees of freedom where required
4Relative binding free energy calculation, and absolute calculations where the series demands it
5Convergence diagnostics: overlap, hysteresis, cycle closure error and replicate agreement
6Retrospective validation against measured affinities before prospective predictions are issued
What We Need From You
1High-quality target structure, preferably a co-crystal with a reference ligand of the series
2The congeneric series with consistent, confirmed binding mode across members
3Measured affinities for a validation subset, which is effectively a prerequisite for a credible study
4The specific transformations you intend, or the design question that should drive the network
What You Receive
1Predicted relative binding free energies with statistical uncertainty per transformation
2Perturbation network diagram with cycle closure errors shown
3Retrospective validation results establishing the expected error on your system
4Ranked compound recommendations with an explicit statement of which differences are significant
5Full protocol and input record enabling the calculation to be repeated or extended
What We Compute and Report
1Relative binding free energy per transformation, in kcal/mol, with statistical uncertainty
2Cycle closure errors across the perturbation network, exposing individually unreliable edges
3Phase-space overlap and hysteresis diagnostics per lambda window
4Retrospective accuracy on your measured compounds, establishing the expected error for your system
5Ranked prospective predictions with the differences that are and are not statistically meaningful
Questions We Are Asked Before Scoping
When is FEP not appropriate?
▸When binding mode differs across the series, when the scaffold changes substantially, when protonation states shift on binding, or when large protein rearrangements accompany binding. We check all of these before accepting the project.
How large a change can be calculated?
▸Small to moderate substituent changes are reliable. Core replacements and charge changes are much harder, need special handling and carry larger errors, which we report per transformation.
Do we need experimental data to start?
▸Strongly recommended. Without a validation set, there is no way to establish the error on your system, and predictions must then be treated as substantially less certain.
Interpretation and Limitations
Computational predictions prioritise hypotheses; they do not independently establish binding, biological activity, therapeutic efficacy, safety or clinical validity. Performance depends on input quality, target class, the structural evidence available and the applicability domain of the selected methods. Scores used for ranking are comparative devices and should not be read as measured quantities. All deliverables are for research use only and are not intended for diagnostic or therapeutic use.
Is This the Right Service for You?
Choose this when
A synthesis decision is expensive enough to justify the most accurate available method
A congeneric series shares a confirmed binding mode and some measured affinities exist
Selectivity between a target and a close off-target needs quantifying
Consider another service when
Many compounds need cheap triage before the expensive calculation
MM/GBSA & MM/PBSA Binding Free Energy The scaffold changes substantially between compounds
De Novo Small Molecule Design What Goes Wrong, and How We Handle It
These are the failure modes that produce confident-looking results which do not survive the next experiment. Each one is designed against rather than discovered afterwards.
Running the method on a series whose binding mode is not consistent
Binding mode consistency is checked before the project is accepted, since the calculation assumes it.
Individual failed transformations averaged into the result
Cycle closure, overlap and hysteresis are checked per edge; failing edges are rerun or reported as unreliable.
Prospective predictions issued with no idea of the error on this system
Measured compounds are reproduced first, and the observed error becomes the error bar quoted for the predictions.
A Worked Example
Choosing between two substituent strategies
The situation
Two vectors were available for optimisation, each requiring a different and costly synthetic route.
What the work involved
A perturbation network was designed with cycle redundancy across both vectors, validated retrospectively against eleven measured compounds, then run prospectively on the untested designs.
What came back
Retrospective error on the series was established and reported. One vector was predicted to gain roughly an order of magnitude, the other nothing measurable; the team committed to a single route.
Illustrative of the shape this work takes. Details are generalised, no client or target is identified, and scope and outcome vary with the system in question.
What Published Work Reports
The results below were published by independent groups and are cited to show what these computational approaches have achieved in the literature. They are not ICDMO project results, and outcomes vary substantially between targets, datasets and programmes. We do not promise comparable results for any particular project.
Alchemical free energy calculations reach roughly 1 kcal/mol error
Benchmark studies on congeneric series across multiple drug targets report root-mean-square errors around 1 kcal/mol, with published protocol work reporting prediction errors below approximately 1.3 kcal/mol on test sets of ligand transformations. An error of this size is small enough to separate compounds differing by roughly one order of magnitude in affinity, which is what makes the method useful for deciding which analogue to synthesise.
Protocol and benchmark studies, Scientific Reports, 2019 ↗Accuracy is system-dependent, and retrospective benchmarks flatter it
Reviews of free energy methods in drug discovery note that much of the published evidence is retrospective, and that prospective performance in live optimisation campaigns is less thoroughly documented. This is why we validate against your own measured compounds before issuing prospective predictions, rather than quoting a literature error bar as though it applied to your system.
Review, Free Energy Calculations for Drug Discovery, 2021 ↗Service Workflow
1
Scientific Consultation
We discuss the objective, the decision it supports and the data you hold. Nothing is quoted before we understand what the result has to answer.
2
Feasibility Review
We check that the data, target class and available structural evidence can support the proposed approach, and say plainly where they cannot.
3
Technical Proposal
Scope, methods, deliverables, timeline and price are confirmed in writing, together with the limitations we expect to report.
4
Computation and QC
The analysis runs on pinned tool versions in a containerised environment, with quality checkpoints and method comparison where that is informative.
5
Report and Follow-Up
You receive the ranked results, the documented methods, the limitations and a proposed validation plan, plus support in interpreting them.
Service Process
1
Online Consultation
›
2
Solution Matching
›
3
Service Contract
›
4
Project Execution
›
5
Data Delivery
Specifications & Turnaround
Note: ICDMO delivers computational (in silico) services. Results prioritise hypotheses for testing; they do not by themselves establish binding, biological activity, efficacy or safety. All deliverables are for research use only and are not intended for diagnostic or therapeutic use. Downstream experimental work is scoped separately and executed through qualified partners.
Get a Custom Quote
Our scientific team responds within 24 hours with a detailed technical proposal and pricing tailored to your project.
Contact Us Online ConsultationStandard Deliverables
✓Sequence-verified final product
✓Full QC documentation
✓Certificate of Analysis (CoA)
✓Electronic data files
✓Project completion report
Why ICDMO?
Computational CDMO Focus
Reproducible, Versioned Pipelines
Confidence & Limitations Reported
Experiment-Aware Recommendations
Response Within One Business Day