Accurate ligand binding mode prediction for novel chemical matter

The domain of applicability of structure-based drug design (SBDD) is limited by the difficulty of obtaining a reliable receptor-ligand structure. The dominant method of obtaining a structure (x-ray crystallography) is both expensive and time consuming. In some cases, such as membrane proteins or GPCR’s, obtaining a structure would be difficult if not impossible.

IFD-MD is able to obtain an accurate structure to use in SBDD more quickly and easily than experimental determination.

Expand the domain of applicability of structure-based drug design with IFD-MD

Explore Alternative Holostructure Conformations

IFD-MD utilizes pharmacophore docking to initially place the ligand while ignoring receptor clashes. This allows for conformational sampling of the receptor in the presence of a docked ligand, rather than refining an empty binding-site.

Utilize WScore To Detect Desolvation

IFD-MD incorporates WScore to detect and penalize desolvation of polar groups caused by non-native ligand poses. In the example shown here, this incorrect ligand pose blocks a backbone NH and is therefore penalized.

Returns Explicit MD Waters

Explicit MD waters are incorporated to provide a complete representation of the system. The placement and interaction with water forms an important part of the IFD-MD scoring function.


Benefits of IFD-MD for Drug Discovery

Provide an accurate structure of novel chemical matter bound to the target in order to understand activity cliffs.

Rationalize Structure Activity Relationships

IFD-MD refines the binding site in the presence of a target ligand, enabling homology models to be used in lead optimization with greater confidence.

Enable Use of Homology Models in Drug Design

Known actives which fail to rigidly dock in existing structures can be re-docked with IFD-MD to generate a competent receptor conformation for this and other actives.

Generate Alternative Receptor Conformations for Virtual Screening

Exploit the rich congeneric series and assay data available in a patent when high-quality structural data doesn’t exist to generate an FEP+ validated structure.

Jumpstart Lead Optimization from Patent Data

IFD-MD returns explicit water molecules at 300K as part of the ligand-receptor complex for a complete model of binding.

Predict and Visualize Bound Waters Crucial for Ligand Binding

IFD-MD’s scoring function provides information on ligand stability and contact quality. This provides rationalization to understand why a pose is most favorable.

Compare Alternative Binding Modes With an Intuitive Scoring Function

Generate a ligand-receptor complex from an HTS hit. Using analogue-by-catalogue and FEP+, the complex can be validated for prospective use.

Rationalize HTS Hits

Avoid the need to initiate a new crystallization program of a known off-target bound to chemical matter.

Understand Off-Target Activity

Identify pose-flips or receptor changes responsible for FEP+ outliers.

Explore FEP+ outliers

Overview of IFD-MD

Introduction to IFD-MD
Walkthrough of an IFD-MD Prediction
Example Applications
IFD-MD and Homology Modeling

Steps to create a valid ligand-receptor structure

Select known binder for structure determination
Perform IFD-MD to generate top 2 models for evaluation
Challenge models with retrospective FEP+
Select performant model (R2, RMSE) for prospective work


IFD-MD Results for 415 retrospective cross-docking experiments taken from publically available structures compared to other docking methods
Training & Test Set ResultsRigid Receptor DockingIFDIFD-MD34%30%41%38%52%56%65%68%85%84%95%93%Top pose<2.5A RMSDTrainingTestOne of Top 2 poses <2.5A RMSDTrainingTest
IFD-MD Results for 415 retrospective cross-docking experiments broken up by protein class
KinaseSerine ProteaseNuclear Hormone ReceptorReductasePhosphotasePhosphodiesteraseAspartic Acid ProteaseHydrolaseSecreted ProteinMembrane Adhesion(Ligand Binding Domain)Nuclear ProteinTransport ProteinIon Channel(Ligand Binding Core)ChaperoneGPCR(Thermostabilized)Peptide Binding Protein020406080100120Number of cross-dock pairsNeither of top 2 poses under 2.5A Ligand RMSDOne of top 2 poses under 2.5A Ligand RMSDTop pose under 2.5A Ligand RMSD

In each experiment the binding pose was predicted starting with a holo structure of the target protein bound to a different ligand. The median tanimoto similarity of the holo-structure ligand to the predicted ligand is 0.04; the ligands were not congeneric. In over 90% of cases a pose within 2.5 Å ligand heavy-atom RMSD to the experimentally determined binding pose was within the top 2 poses predicted by IFD-MD. In over 80% of those cases the single top scoring pose was within 2.5 Å RMSD. For comparison the older IFD can identify a pose under 2.5 Å RMSD in the top 2 in only 60% of these cases and in the single top ranked pose in only 50% of cases.


Blog Post


Dr. Ed Miller describes how next generation induced fit docking workflows can provides an accurate way to determine binding modes of novel ligand scaffolds.


Science Article

IFD-MD App Note

IFD-MD takes the available structure and predicts atomic details of the protein-ligand complex structure needed for SBDD. See the details of how it performs in this application note.


Blog Post

STORIES FROM DRUG DISCOVERY: Modeling Strategies in the Pursuit of Development Candidate in Oncology Program 1

Dr. Sayan Mondal, a Research Leader in Schrödinger’s Drug Discovery Group, demonstrates modeling strategies used from start to finish in a two-year oncology project.


Training Materials

Cross-Docking with IFD-MD

In this tutorial you will learn how to use IFD-MD to generate a predicted binding pose of a known active compound using a holo crystal structure solved with a different ligand as a starting point.

Expanding the Domain of Applicability with IFD-MD Webinar

An overview of IFD-MD and presentation of the results of 41 targets consisting of 415 cross-docks divided among a training and test set.


Induced-Fit Docking Enables Accurate Free Energy Perturbation Calculations in Homology Models

Tianchuan Xu, Kai Zhu, Alexandre Beautrait, Jeremie Vendome, Kenneth Borrelli, Robert Abel, Richard Friesner, Edward Miller 

2022 View
Reliable and Accurate Solution to the Induced Fit Docking Problem for Protein-Ligand Binding

Miller, E.; Murphy, R.; Sindhikara, D.; Borrelli, K.; Grisewood, M.; Ranalli, F.; Dixon, S.; Jerome, S.; Boyles, N.; Day, T.; Ghanakota, P.; Mondal, S.; Rafi, S.B.; Troast, D.M.; Abel, R.; Friesner, R.A.,

J. Chem. Theory Comput., 2021 View


Expand the domain of applicability of structure-based drug design with IFD-MD



Back To Top