Muammar El Khatib

Muammar El Khatib

Principal Scientist

Odyssey Therapeutics

Biography

Muammar is a Principal Scientist in Chemical Informatics and Machine Learning applying data-driven models to solve problems in drug discovery at Odyssey Therapeutics. His research interests include machine learning, physical chemistry, and software solutions.

He is a chemist by training from the University of Zulia in Venezuela and started his graduate studies with a European Master in Theoretical Chemistry and Computational modeling of the Erasmus Mundus Program. His Ph.D. in theoretical chemical physics was about the characterization of metallic and insulating properties of low-dimensional systems using the theory of the insulating state of Walter Kohn applied with wave function theory.

He was a postdoctoral research associate at Brown University, where he worked in the acceleration of atomistic simulations with machine learning models in the group of Prof. Andrew A. Peterson in the Catalyst Design Laboratory. He acquired experience with neural networks and kernel ridge regression models to mimic quantum mechanics simulations using interatomic machine learning potentials in this appointment.

At Lawerence Berkeley National Laboratory, he was a postdoctoral scholar working towards the development of machine learning approaches, algorithms and data sets to solve chemical science problems.

His first work in the pharmaceutical field started at Bristol-Myers Squibb where he applied machine learning to the prediction of physicochemical properties to different assays of interest systematically studying a plethora of algorithms. He also introduced and championed the addition of uncertain quantification as a good practice in predictive models used for decision-making in the company, and multi-task learning to build more robust methods.

He has published more than ten papers, and given presentations at international conferences, and developed the ML4Chem machine learning package, a module for the MOLPRO quantum-chemistry package, and the atomistic machine-learning package (Amp). Additionally, he has participated in the free software community and is a Debian Linux developer.

Interests
  • Machine Learning (deep and shallow).
  • Uncertainty Quantification.
  • Software Development.
  • Theoretical Chemical Physics.
  • Electronic Structure.
Education
  • PhD in Theoretical Chemical Physics, 2015

    Université Paul Sabatier, France

  • Master in Theoretical Chemistry and Computational Modeling, 2012

    Université Paul Sabatier, France

  • BSc in Chemistry, 2010

    University of Zulia, Venezuela

Skills

Technical
Code
Data Science
Physical Chemistry
Hobbies
Kung Fu
Guitar
Technology

Experience

 
 
 
 
 
Principal Scientist Chemical Informatics and Machine Learning
April 2024 – Present Boston, MA, United States

Responsibilities include:

  • Drive strategic implementation of chemical informatics and AI/ML techniques to expedite drug discovery guided by the needs of teams to advance Discovery assets to the clinic.
  • Champion the integration of chemical informatics with proteomic data, utilizing three-dimensional molecular modeling to elucidate precise mechanisms of protein-ligand interactions, critical for target validation and lead optimization.
  • Orchestrate the synthesis and analysis of multidimensional datasets, with a focus on integrating 3D structural insights with chemical properties to pioneer predictive models for molecular behavior and drug efficacy.
  • Develop and maintain cutting-edge frameworks for the semantic integration and intelligent retrieval of multidimensional chemical and biological data.
  • Facilitate and expand strategic partnerships with leading data vendors, academic pioneers, and technology innovators to advance Odyssey’s capabilities in cheminformatics and computational drug discovery.
  • Act as a pivotal thought leader, demystifying the complexities of cheminformatics and machine learning for cross-functional teams, fostering a culture of innovation and knowledge sharing.
 
 
 
 
 
Principal Scientist
November 2020 – March 2024 Cambridge, MA, United States

Responsibilities include:

  • Report to the lead for Predictive Sciences of the IPS organization; scenarios involve a range of datasets and learning objectives, including drug discovery, multi-modal modeling, uncertainty quantification, and predictive models for chemical and biological datasets.
    • Formulation and implementation of predictive modeling and machine learning solutions for optimizing chemical structures, drug-target interactions, and physicochemical properties.
    • Application of cutting-edge machine learning (deep learning) approaches to physicochemical property predictions and molecular interaction challenges.
    • Work alongside experts in similar applications of machine learning in the biotechnology domain, including:
      1. Collaboration to develop human-in-the-loop systems to capture and operationalize machine learning datasets and algorithms used by BMS scientists.
      2. Application of supervised and semi-supervised deep learning methods to derive robust, generalizable, and reusable representations for chemical and biological assay data.
      3. Design of multi-task, multi-modal, and generative neural network learning approaches to tackle real-world drug discovery optimization problems.
    • Pursue leading research in applied scientific machine learning that demonstrates the value of predictive methods to accelerate and optimize drug development.
    • Derive and apply predictive approaches in collaboration with BMS colleagues in the Informatics Predictive Sciences and Chemistry departments.
    • Apply rigorous internal standards for applied machine learning practice, including evaluation of methods, approaches, and solutions.
    • Present strategies, techniques, results, and conclusions to BMS colleagues and external audiences.
    • Enable strategic collaborations with academic and commercial collaborators to benefit therapeutic programs. I am the BMS representative before the “Machine Learning for Pharmaceutical Discovery and Synthesis Consortium” collaboration with the Massachusetts Institute of Technology. For more information, see https://mlpds.mit.edu/.
    • Implement and apply uncertainty quantification of predictive models to steer scientific discovery.
    • Supervise and mentor interns to develop their careers and benefit therapeutic programs at BMS.
 
 
 
 
 
Postdoctoral Scholar
November 2018 – September 2020 Berkeley, CA, United States

Responsibilities include:

  • Developed a Python library to ease the deployment of machine learning models for chemistry and materials sciences. https://github.com/muammar/ml4chem. This package is helping us advance our research faster because we can consistently implement new methods.
  • Developed a neural network model that can learn how to predict retention times from chromatography data. These models can be used by experimentalists to get insights about the substances they study without explicitly running the experiments.
  • Applied autoencoders to extract features and systematically studied their topology to understand their effect on the predictive power of models used for material sciences.
  • Generated data sets using web-scraping and diversified their variance with active learning techniques.
  • Worked on a project for Scaling Interactive Science for Data-Intensive Discovery for the Linac Coherent Light Source (LCLS) at SLAC National Accelerator Laboratory.
  • Led and designed research projects executed by summer interns.
  • Wrote scientific publications to show our results to the community.
 
 
 
 
 
Postdoctoral Research Associate
October 2016 – November 2018 Providence, RI, United States

Responsibilities include:

  • Worked on the acceleration of electronic structure calculations using machine learning models to decrease orders of magnitude of the computational time needed by the simulations.
  • Was actively involved in the development of the Atomistic Machine-learning Package (Amp) created and maintained by the Catalyst Design Laboratory at Brown University https://bitbucket.org/andrewpeterson/amp.
  • Implemented kernel ridge regression within an atom-centered mode in their machine-learning package.
  • Participated in the design of scientific projects and supervision of students during their research in our laboratory.
  • Presented scientific results in international conferences.
 
 
 
 
 
PhD candidate in Theoretical Chemical Physics
July 2012 – July 2015 Toulouse, France
My Ph.D. thesis was titled: “Characterization of metallic and insulating properties of low-dimensional systems.”
 
 
 
 
 
Master in Theoretical Chemistry and Computational Modeling
Université Paul Sabatier, Università degli Studi di Perugia
September 2010 – July 2012 Toulouse, France and Perugia, Italy
I was awarded an Erasmus Mundus Scholarship to do my studies in different European universities. My dissertation was titled: “An ab initio potential energy surface for quantum reactive scattering calculations.”
 
 
 
 
 
B.S. in Chemistry
May 2010 – October 2001 Maracaibo, Venezuela
My dissertation was titled: “Determination of the Linear and Nonlinear Optical Properties of the Nitrogenous Bases of DNA, RNA and their respective tautomers.”

Recent Publications

Quickly discover relevant content by filtering publications.
(2020). Open Chemistry, JupyterLab, REST, and quantum chemistry. Int. J. Quantum Chem..

PDF Cite Code

Contact

I provide two ways you can contact me.