Khashan LaboratoryMolecular Modeling & Simulation

Khashan Laboratory · Brooklyn, New York

Molecular Modeling & Simulation Laboratory

We build computational methods that help explain how molecules interact, how proteins work, and how better drug-design ideas can be found.

Explore our research See research software

What we do
Computer-aided drug design, molecular dynamics simulation, and efficient algorithms for cheminformatics and bioinformatics.

Our approach
Turn molecular structure into useful evidence: discover patterns, test models, and explain biological mechanisms.

Introduction

From molecular patterns to practical insight

Our work brings computational chemistry, structural biology, and algorithm development together to answer questions that are difficult to see experimentally.

The laboratory studies the geometry and chemistry of molecular structures, especially the interfaces where proteins and small molecules meet. By finding recurring patterns in these structures, we develop tools for predicting interactions, guiding drug discovery, and understanding biological function.

01

Cheminformatics

Molecular descriptors, pharmacophoric patterns, fragment libraries, and structure-based screening.

Explore the area →
02

Bioinformatics

Structural motifs that help assess protein interfaces, folds, and potential biological function.

Explore the area →
03

Molecular dynamics

Simulation to reveal how binding changes molecular motion and influences signaling.

Explore the area →

Research

A structural view of biology

We represent structures as graphs, mine common subgraphs, and use recurring geometric and chemical features to build predictive methods.

Research begins with graph representations of native molecular structures, macromolecules, or the interfaces between them. Efficient subgraph mining identifies frequent structural features, or motifs, that can be used to predict biomolecular structure and function.

In cheminformatics, this approach has supported molecular descriptors, common pharmacophoric groups, fragment-based virtual libraries, and interaction patterns for assessing small-molecule docking. The laboratory also develops docking scores that address limitations caused by missing entropy contributions.

In bioinformatics, frequent geometric motifs have been used to assess protein-protein docking decoys, identify correct protein folds, and investigate structural features associated with protein function. Molecular dynamics complements this work by examining binding processes and the conformational changes that may trigger signal transduction.

Research area

Cheminformatics

Computational methods for molecular descriptors, QSAR/QSPR modeling, pharmacophore discovery, fragment-based virtual screening, and protein-ligand interaction analysis.

Research area

Bioinformatics

Mining structural patterns in protein interfaces and internal residues to assess docking poses, protein folds, and structure-function relationships.

Research area

Molecular dynamics simulation

Large-scale MD simulations explore molecular binding, conformational change, and the mechanisms that connect molecular events to biological signaling.

Ahead of the AI boom

The methods were already taking shape

Today’s generative-AI moment did not create computational discovery. It made a much older body of work newly visible.

A long arc—not a sudden pivot

Long before generative AI became mainstream, Ra’ed Khashan was turning molecular structures into machine-readable graphs, mining recurring chemical patterns, and building computational systems to rank, predict, and design molecular interactions.

01 · Representation

Make chemistry computable

Molecules and interfaces become labeled graphs whose atoms, residues, geometry, and chemistry can be analyzed by algorithms.

02 · Pattern discovery

Learn what repeats

Frequent subgraph mining finds fragments and multi-body interaction motifs that recur across known molecular structures.

03 · Prediction

Use patterns to decide

QSAR models, docking scores, pocket searches, and neural networks turn those representations into rankings and testable candidates.

2003

Molecular descriptors

Refining BCUT descriptors for computer-assisted drug discovery.

2005–07

Graph mining

Fragment-based descriptors from frequent common subgraphs; PhD work in computational drug discovery.

2012

Interface intelligence

SPIDER and FragVLib convert structural patterns and pocket similarity into usable methods.

2014

Predictive features

Published fragment-based chemical descriptors applied to QSAR modeling.

2022 →

Neural & AI-assisted design

ANN-based MBI-Score and the FragDeNovo concept extend the same foundation with machine learning.

Research, made visible

Four ways structure becomes knowledge

Conceptual illustrations translate the laboratory’s original figures into a clearer visual language for students, collaborators, and non-specialists.

A protein-ligand interface resolving into a graph of recurring molecular interaction patterns
Graph mining

From structure to recurring pattern

Molecular structures are represented as graphs. Repeated geometric and chemical subgraphs become interpretable features for screening, modeling, and scoring.

Two protein surfaces joined at an interface with selected multi-residue geometric motifs
SPIDER

Recognizing native-like interfaces

SPIDER compares the geometry of residues at a proposed protein interface with interaction patterns learned from naturally occurring complexes.

Conceptual sequence of an RNA-guided protein complex recognizing DNA and stabilizing an RNA-DNA hybrid
Molecular dynamics

Following molecular change over time

Simulation helps investigate recognition, conformational change, and RNA-guided DNA targeting at a level that can be difficult to observe experimentally.

Molecular fragments being assembled into candidate ligands and evaluated against a protein pocket
FragDeNovo

Building candidates from compatible fragments

Pocket similarity retrieves useful molecular fragments; computational design and machine learning then help propose larger candidates for further evaluation.

Energy landscape comparing a broad conformational state A with a deeper but narrower state B
Energy & entropy

Why the deepest well is not the whole story

A scoring method that considers potential energy alone may favor state B. The broader basin of state A represents more accessible conformations, illustrating why entropy can change the free-energy picture.

These illustrations explain research concepts; they are not experimental results, measured structures, or claims of clinical performance.

Research software

Methods made usable

Tools developed around recurring geometric and chemical features in protein structures and molecular interfaces.

A protein-ligand interface becoming a graph of recurring interaction patterns

MBI-Score

Docking

An artificial-neural-network scoring function for protein-ligand docking, built from frequent geometric and chemical patterns at native interfaces.

In plain language: learn how atoms interact in known complexes, then use those patterns to rank possible binding poses.

Two protein surfaces connected by selected multi-residue interface patterns

SPIDER

Protein interfaces

A multi-body interaction scoring method that uses patterns of exposed interfacial residues to rank protein-protein docking decoys.

In plain language: compare how proteins touch, then move the most natural-looking pairings toward the top.

Similar protein pockets leading to an organized collection of compatible molecular fragments

FragVLib

Fragment design

A free program for generating fragment-based virtual libraries by searching ligand-receptor pockets for structural similarity.

In plain language: find pockets shaped like the target and collect molecular pieces that already fit similar spaces.

Many molecular fragments passing through shape and interaction filters toward a target pocket

FragVScreen

Virtual screening

A fragment-based virtual screening approach that combines pocket similarity with geometric and chemical interaction patterns.

In plain language: filter many fragments down to the few that best match both the pocket’s shape and chemistry.

Compatible molecular fragments being assembled into candidate ligands for a protein pocket

FragDeNovo

De novo design

AI-supported fragment-based de novo design using pocket similarity search across native protein-ligand complexes.

In plain language: combine promising fragments into new candidate molecules for computational evaluation and further testing.

Many protein structures with highlighted cavities becoming an organized pocket collection

PDBPockets

Structural data

An automated analysis of more than 220,000 Protein Data Bank structures using Fpocket. The dataset includes predicted ligandable pockets, pocket geometry, and residue-level annotations for structure-based discovery.

In plain language: scan protein structures for possible binding sites and organize the results into searchable data.

Chemical, target, biological, and safety information converging into unified drug records

Drug Dataset

Data integration

A reproducible pipeline that brings chemical, target-binding, therapeutic, and safety information together for nearly 6,000 small-molecule drugs in a machine-learning-ready dataset.

In plain language: bring scattered facts about each drug into one consistent record that computers can analyze.

Two related protein pockets showing an exchanged molecular group with a similar fit

ChemIsosteres

Cheminformatics

A structure-based approach to finding bioisosteres that share related binding pockets and similar geometric and chemical features.

In plain language: suggest a different chemical group that may preserve a molecule’s shape, interactions, and pocket fit.

Proposed research direction

The next frontier

The Pocket Intelligence Engine

An explainable-AI concept that connects the laboratory’s structural datasets, scoring methods, and molecular-design tools into one discovery pipeline.

What if a protein pocket could be searched as easily as a sequence—and every match could lead to a drug, an off-target warning, or a new molecule worth testing?

01 · Map

Represent the pocket

Convert pocket shape, residues, geometry, and chemical character into a machine-readable structural fingerprint.

PDBPockets
02 · Learn

Find hidden relatives

Train an explainable model to recognize similar binding environments—even across proteins with unrelated sequences.

Pocket similarity + ML
03 · Connect

Link pockets to drugs

Search integrated drug, target, therapeutic, and safety records for molecules already associated with matching pockets.

Drug Dataset
04 · Design

Improve the candidate

Retrieve compatible fragments, propose replacements, and assemble new candidates while preserving useful interactions.

FragVScreen + ChemIsosteres
05 · Explain

Rank with evidence

Score each pose and show which geometric and chemical interaction patterns support the prediction.

MBI-Score
Opportunity 01

Drug repurposing

Reveal an unexpected pocket match that suggests an existing drug could be investigated for a different target or disease.

Opportunity 02

Off-target foresight

Identify similar pockets elsewhere in the proteome that may help explain side effects or guide safer molecular design.

Opportunity 03

Explainable design

Generate new fragments or substitutions while showing the structural evidence behind every recommendation.

A path from idea to evidence

Make the first claim testable

The first contribution would not need to be a new medicine. A rigorous benchmark and one experimentally testable prediction could establish the platform’s value.

  1. Build a transparent baseline. Create pocket fingerprints and test whether known drug–target relationships are recovered.
  2. Challenge the model. Separate proteins by family and time so the evaluation measures generalization, not memorization.
  3. Rank novel hypotheses. Prioritize candidates using pocket similarity, docking, interaction patterns, and safety evidence.
  4. Validate one prediction. Share a reproducible computational result and pursue an experimental collaboration for confirmation.

Status: This is a proposed research direction inspired by the laboratory’s published methods and public datasets—not a completed system, validated discovery, or claim of therapeutic performance. Explore the PDBPockets publication ↗

Publications

Research record

Verified recent journal articles together with conference abstracts, theses, book chapters, and works preserved from the laboratory’s original publication list.

View the complete publication list (30 entries)
  1. Hinal Gajjar and Ra’ed Khashan. PDBPockets: A Dataset of Potential Ligand-Binding Pockets Identified with Fpocket and Prepared for Large-Scale Inverse Docking. Computational Biology and Chemistry, 123, 109007, 2026. doi: 10.1016/j.compbiolchem.2026.109007.
  2. Ali Khudhir, Mahmoud A. Al-Sha’er, Mahmoud A. Alelaimat, and Raed Khashan. Identification and Biological Assessment of Amino Benzoxazole Derivatives as KDR Inhibitors and Potential Anti-Cancer Agents. Journal of Computer-Aided Molecular Design, 39(2), 93, 2025. doi: 10.1007/s10822-025-00665-6.
  3. Abdulraouf Ramadan, Pushpa Rao, Saleh Allababidi, Raed Khashan, and Anas M. Fathallah. Tolerization with a Novel Dual-Acting Liposomal Tim Agonist Prepares the Immune System for the Success of Gene Therapy. International Journal of Molecular Sciences, 26(8), 3830, 2025. doi: 10.3390/ijms26083830.
  4. Amnah Alalmaie and Raed Khashan. Mechanistic Insight Into the Conformational Changes of Cas8 Upon Binding to Different PAM Sequences in the Transposon-Encoded Type I-F CRISPR-Cas System. Proteins: Structure, Function, and Bioinformatics, 92(12), 1428–1448, 2024. doi: 10.1002/prot.26730.
  5. Rwan Elsanhoury, Abdulaziz Alasmari, Prashanth Parupathi, Mouhannad Jumaa, Suliman Al-Fayoumi, Avinash Kumar, Raed Khashan, Sami Nazzal, and Ahmed Abu Fayyad. AI & Experimental-Based Discovery and Preclinical IND-Enabling Studies of Selective BMX Inhibitors for Development of Cancer Therapeutics. International Journal of Pharmaceutics, 645, 123384, 2023. doi: 10.1016/j.ijpharm.2023.123384.
  6. Amnah Alalmaei, Saousen Diaf, and Raed Khashan. Insight into the Molecular Mechanism of the Transposon-Encoded Type I-F CRISPR-Cas System. Journal of Genetic Engineering and Biotechnology, 21(1), 60, 2023. doi: 10.1186/s43141-023-00507-8.
  7. Raed Khashan. SCORPIONS: A Scoring Function for Computationally Generated Protein Folds using 3D Structural Motifs derived by Mining Internal Residues in Native Protein Structures. In preparation.
  8. Raed Khashan and Grace Brannigan. Estimating Binding Affinities of GABAA Neuromodulators: Using Free Energy Methods in Combination with Large Scale Molecular Dynamics Simulations. In preparation.
  9. Raed Khashan and Ingo Muegge. mbiPMF: Accounting for Multi-body Interactions in the Knowledge-based PMF Scoring Function for Protein-Ligand Interactions. In preparation.
  10. Raed Khashan. ChemIsosteres: Structure-based Bioisosteres Sharing Related Binding Pockets with Similar Geometric & Chemical Features. In preparation.
  11. Raed Khashan and Faruk Khan. Chapter 35: Drug Discovery and Development. In Medicinal Chemistry for Pharmacy Students, edited by M. O. Faruk Khan. In preparation.
  12. Saousen Diaf, Amnah Alalmaei, Sanzana Rivu, and Raed Khashan. Implications of Molecular and Structural Mechanism of Insulin Receptor on Receptor Activation and Drug Design. Membranes (under review).
  13. Raed Khashan. Using AI Strategies to Support Fragment-based De Novo Drug Design (FragDeNovo) using Pocket Similarity Search of Native Protein-Ligand Complexes. Molecular Informatics (under review).
  14. Raed Khashan. Data Mining Meets Machine Learning: Performing Fragment-based Virtual Screening (FragVScreen) using Pocket Similarity Search of Native Protein-Ligand Complexes. Journal of Chemical Information and Modeling (under review).
  15. Raed Khashan, Alexander Tropsha, and Weifan Zheng. Data Mining Meets Machine Learning: A Novel ANN-based Multi-Body Interaction Docking Scoring Function (MBI-Score) based on Utilizing Frequent Geometric and Chemical Patterns of Interfacial Atoms in Native Protein-Ligand Complexes. Molecular Informatics, 41(8), e2100248, 2022. doi: 10.1002/minf.202100248.
  16. Raed Khashan. Chapter 3: Generating “Fragment-Based Virtual Library” Using Pocket Similarity Search of Ligand-Receptor Complexes. In Fragment-Based Methods in Drug Discovery, Methods in Molecular Biology, vol. 1289, pp. 23–30, 2015.
  17. Raed Khashan, Weifan Zheng, and Alexander Tropsha. The Development of Novel Chemical Fragment-Based Descriptors Using Frequent Common Subgraph Mining Approach and Their Application in QSAR Modeling. Molecular Informatics, 33(3), 201–215, 2014.
  18. Raed Khashan. FragVLib – A Free Program for Generating “Fragment-based Virtual Library” Using Pocket Similarity Search of Ligand-Receptor Complexes. Journal of Cheminformatics, 4(1), 18, 2012.
  19. Raed Khashan, Weifan Zheng, and Alexander Tropsha. Scoring Protein Interaction Decoys using Exposed Residues (SPIDER): A Novel Multi-Body Interaction Scoring Function based on Frequent Geometric Patterns of Interfacial Residues. Proteins: Structure, Function, and Bioinformatics, 80(9), 2207–2217, 2012.
  20. Sarel J. Fleishman, Timothy A. Whitehead, Raed Khashan, Stephen Bush, Denis Fouches, Alexander Tropsha, et al. Community-Wide Assessment of Protein-Interface Modeling Suggests Improvements to Design Methodology. Journal of Molecular Biology, 414(2), 289–302, 2011.
  21. Raed Khashan, Weifan Zheng, and Alexander Tropsha. FragVLib: Fragment-based virtual screening library using geometric and chemical patterns of interactions at interface of ligand-receptor complex crystal structures. Abstracts of Papers, 240th ACS National Meeting, Boston, MA, August 22–26, 2010, COMP-216.
  22. Raed Khashan, Weifan Zheng, and Alexander Tropsha. GeoIsosteres: Structure-based approach to finding bioisosteres using geometric & chemical patterns of interacting atoms at receptor-ligand interfaces. Abstracts of Papers, 240th ACS National Meeting, Boston, MA, August 22, 2010, COMP-414.
  23. Raed Khashan, Weifan Zheng, and Alexander Tropsha. Fragment based design and biophores using geometric and chemical patterns of interactions at interface of ligand-receptor complex crystal structures. Abstracts of Papers, 240th ACS National Meeting, Boston, MA, August 22, 2010, COMP-275.
  24. Raed Khashan, Weifan Zheng, Wei Wang, and Alexander Tropsha. Development of scoring functions for protein ligand binding based on frequent geometric and chemical patterns of inter-atomic interactions at their interfaces. Abstracts of Papers, 234th ACS National Meeting, Boston, MA, August 19–23, 2007, COMP-352.
  25. Raed Khashan, Weifan Zheng, Wei Wang, and Alexander Tropsha. Development of docking protocols & scoring functions using frequent geometric & chemical patterns of inter-atomic interactions at the interface of protein-ligand complexes. Abstracts of Papers, 233rd ACS National Meeting, Chicago, IL, March 25–29, 2007, COMP-191 & COMP-261.
  26. Raed Khashan. Development and Application of Ligand/Structure-based Computational Drug Discovery Tools Based on Frequent Subgraph Mining of Chemical Structures. Ph.D. dissertation, University of North Carolina at Chapel Hill, August 2007.
  27. Raed Khashan, Weifan Zheng, Jun Huan, Wei Wang, and Alexander Tropsha. Development of fragment-based chemical descriptors using novel frequent common subgraph mining approach and their application in QSAR modeling. Abstracts of Papers, 230th ACS National Meeting, Washington, DC, August 28–September 1, 2005, COMP-177 & COMP-337.
  28. Scott Oloff, Raed Khashan, Robert Plourde, and Alex Tropsha. Development of validated QSAR models of P2Y12 receptor antagonists and their application to database mining. Abstracts of Papers, 227th ACS National Meeting, Anaheim, CA, March 28–April 1, 2004, MEDI-300.
  29. Raed Khashan. Refinement and Validation of BCUT Descriptors for Computer Assisted Drug Discovery. M.S. thesis, University of Texas at Austin, August 2003.
  30. Robert Pearlman, Raed Khashan, D. Wong, and Renzo Balducci. ProtoPlex: User-control over tautomeric & protonation state. Abstracts of Papers, 224th ACS National Meeting, Boston, MA, August 18–22, 2002, COMP-232.

About the PI

Ra’ed S. A. Khashan, PhD, RPh

Associate Professor of Pharmaceutical Sciences at Long Island University, computational medicinal chemist, educator, licensed pharmacist, and developer of methods for understanding molecular interactions.

Ra’ed Khashan
Associate Professor · Principal Investigator
Education

PhD, Pharmaceutical Sciences (Medicinal Chemistry) · University of North Carolina at Chapel Hill, 2003–2007. Dissertation: “Development & Application of Ligand/Structure-based Computational Drug Discovery Tools Based on Frequent Subgraph Mining of Chemical Structures.” Advisor: Prof. Alexander Tropsha.

MS, Pharmaceutical Sciences (Physical Pharmacy) · The University of Texas at Austin, 2000–2003. Thesis: “Refinement and Validation of BCUT Descriptors for Computer Assisted Drug Discovery.” Advisor: Prof. Robert S. Pearlman.

BS, Computer Science (minor in Chemistry) · Yarmouk University, 1997–2000; Bachelor of Pharmacy · Jordan University of Science & Technology, 1994–1999. Both degrees listed with honors.

Academic and research experience

Associate Professor, Division of Pharmaceutical Sciences, Long Island University–Brooklyn · 2022–present. Research includes transposon-encoded Type I-F CRISPR/Cas DNA targeting, insulin-receptor mechanisms, and AI-supported fragment-based de novo drug design.

Assistant/Associate Professor, USciences, Philadelphia · 2018–2022. Work included CRISPR/Cas and insulin-receptor mechanisms, plus an ANN-based multi-body interaction docking score.

Assistant Professor (Founding Faculty), UT Tyler · 2014–2017. Developed SCORPIONS for protein-fold scoring and methods for classifying protein families and functions from structural patterns.

Assistant Professor, King Faisal University · 2012–2014. Developed pocket-similarity methods for FragVLib and FragVScreen, structure-based bioisostere discovery, and multi-body interaction scoring.

Research Fellow, North Carolina Central University · 2008–2011. Developed MBI scoring methods and SPIDER for protein-protein interface decoys.

Earlier appointments include visiting scholar at UNC Chapel Hill, research fellow at Rutgers University–Camden, research assistant at UNC Chapel Hill and UT Austin, and an internship at Inspire Pharmaceuticals.

Research interests

Computer-aided drug design; molecular dynamics simulation; cheminformatics and bioinformatics; fragment-based design; molecular descriptors; docking and scoring; graph mining of geometric and chemical interaction patterns.

Teaching, pharmacy, and mentorship

Dr. Khashan brings more than two decades of computational research and education to undergraduate, graduate, and PharmD teaching. He has mentored doctoral and professional students and connects computational theory to patient-centered practice through more than ten years of community-pharmacy experience.

Selected recognition
  • Research & Scholar Award, Philadelphia College of Pharmacy, 2020
  • AACP Teacher of the Year Award, 2015
  • Distinguished Faculty Member Award, King Faisal University, 2012
  • ACS COMP Division Chemical Computing Group Excellence Award, 2007
  • Excellence in Scholarship and Graduate School Dissertation Fellow, UNC Chapel Hill, 2006
Professional leadership and verified profile

Founder and CEO of Artelligence Therapeutics LLC, as disclosed in a peer-reviewed 2025 publication.

View the official Long Island University faculty profile ↗
View the public authorship and company disclosure ↗

Contact

Start a conversation

For research questions, collaborations, or information about the laboratory, get in touch.

Address
1 University Plaza
Brooklyn, NY 11201-8423
Institutional email
Raed.Khashan@liu.edu
Laboratory
Khashan Laboratory
Long Island University–Brooklyn