SCC78 2024

Posted on November 8, 2024November 8, 2024 by Anonymous

Conference

SCC78 2024

Date & Time

December 11th-13th, 2024

Location

Los Angeles, California

Schrödinger is excited to be participating in the SCC78 conference taking place on December 11th – 13th in Los Angeles, California. Join us for a presentation by Haidong Liu, Senior Scientist at Schrödinger, titled “Screening Antioxidant Ingredients Using Machine Learning and Physics-based Modeling .”

Presentation

DEC 12 | 3:30 PM

Session G: AI Beauty Revolution

Screening Antioxidant Ingredients Using Machine Learning and Physics-based Modeling

Speaker:
Haidong Liu, Senior Scientist, Schrödinger

Abstract:
Antioxidants are an important ingredient for cosmetic products to alleviate oxidative stress. While high-throughput screening for new antioxidant candidates still remains challenging experimentally. And the data-driven machine learning models would require the input of a reliable dataset. Here we present an efficient computational approach that combines the physics-based and machine learning tools to address this issue, and this approach only uses molecular structures as inputs.

We used molecular quantum mechanical (QM) calculation and machine learning to predict the antioxidant activity through hydrogen atom transfer (HAT) mechanism. We first constructed a library of flavonoid structures and then calculated the hydrogen dissociation energies of the hydroxyl group in solvents using QM. The machine learning model was trained and validated using the hydrogen dissociation energies from QM calculations. We can easily screen thousands of molecules, and this physics-based and machine learning combined approach can be used for other properties.

MRS Fall 2024

Posted on October 31, 2024November 4, 2024 by Anonymous

Conference

MRS Fall 2024

Date & Time

December 1st-6th, 2024

Location

Boston, Massachusetts

Schrödinger is excited to be participating in the MRS Fall 2024 conference taking place on December 1st – 6th in Boston, Massachusetts. Join us for presentations by Schrödinger scientists on Dec 4th and 5th. Additionally, attend a presentation on Dec 3rd by Panasonic, co-authored by Schrödinger, titled “Discovering Low-Viscosity Molecules Using an Integrated Physics-Based Modeling, High-Throughput Screening, and Active Learning Approach (2)— Screening from PubChem Database.”

Presentations

DEC 3 | 8:00 PM

Hynes, Level 1, Hall A

Discovering Low-Viscosity Molecules Using an Integrated Physics-Based Modeling, High-Throughput Screening, and Active Learning Approach (2)— Screening from PubChem Database

Presenters:
Nobuyuki Matsuzawa, Hiroyuki Maeshima, Tatsuhito Ando, Atif Afzal, Benjamin Coscia, Andrea Browning, Mathew Halls, Karl Leswing, Tsuguo Morisato

Schrödinger collaborated with Panasonic on this presentation

DEC 4 | 3:45 PM

Hynes, Level 3, Ballroom C

Discovering Low-Viscosity Molecules Using an Integrated Physics-Based Modeling, High-Throughput Screening and Active Learning Approach (1)— Screening from the GDB Database

Speaker:
Atif Afzal, Principal Scientist

Abstract:
The discovery of low-viscosity molecules is crucial for the development of next-generation batteries and capacitors. Large molecular libraries available in the literature provide a valuable resource for identifying promising candidates. In this study, we utilized the GDB database1, one of the largest repositories of small molecules, to identify low-viscosity molecules. We employed and benchmarked molecular dynamics methods to accurately compute the dynamic properties without the need for synthesis or empirical testing, validating our calculations against experimental data. However, the number of molecules of interest from the GDB database is too large (several hundreds of thousands), making it impractical to identify promising candidates using purely physics-based models due to computational costs. Therefore, we implemented advanced machine learning (ML) techniques and smart selection approaches to dramatically reduce the number of physics-based calculations needed. Physics-based simulations of viscosity included both Green-Kubo and Einstein-Helfand approaches allowing for robust calculation across the selected molecules. By employing an active learning approach, we optimized the selection of molecules, enhancing the efficiency of the ML model while targeting low-viscosity candidates. Additionally, we computed the boiling points (BP) of the molecules using ML models trained on experimental BP data. As a result, we identified more than 100 molecules with viscosities less than 0.35 cP and BP above 80°C. We demonstrate that by integrating accurate physics-based models with advanced ML techniques, we can effectively identify top molecular candidates while significantly reducing computational costs.

DEC 5 | 11:15 AM

Sheraton, Second Floor, Constitution B

Prediction of aqueous and non-aqueous solubility using machine learning

Speaker:
Lihua Chen, Senior Scientist

Abstract:
Solubility, the capacity of a solute to dissolve in a solvent, forming a solution, is a crucial design parameter across various materials and life science applications. Due to the high cost of experimental measurements, we have developed quantitative structure-property relationship (QSPR) models to rapidly and accurately predict aqueous solubility in water and non-aqueous solubility in organic solvents. For this purpose, we gathered 14,485 room temperature aqueous solubility data points and 45,313 temperature-dependent non-aqueous solubility data points from literature and open-source databases. Additionally, we incorporated advanced cheminformatics-based, graph-based, and physics-based descriptors computed through classical molecular dynamics to optimize machine learning performance. These models can significantly streamline molecular discovery by providing rapid, accurate solubility predictions, reducing the need for costly experiments, and accelerating the identification and optimization of promising candidates.

Release 2024-4

Release Notes

Release 2024-4

Small Molecule Drug Discovery

Platform Environment

Maestro Graphical Interface

Improved usability in scatter plots and histograms:
- See relationships in data across multiple plots and histograms with streamlined menu into “Entry Actions” and “Sync Options” menu icon
- Added support for string and boolean histograms
In the histogram panel, easily switch between settings and data table views
Specify the number of columns and rows for fine control of Workspace Tiles
Save animated GIF of vibrational motion from Jaguar frequency calculation
Enhanced Cryo-EM surface performance:
- Up to 2x faster loading of Cryo-EM surface files
- Up to 5x increase in speed for isosurface contour creation and adjustments
Refined toolbar design for enhanced simplicity, modern aesthetics, and optimization for dark mode
Updated Maestro Project format to version 5:
- Support for multi-letter chain names beyond traditional 26 characters
- Compressed .prjzip files designed for easy sharing via email
- Automatic conversion of version 4 projects to version 5 upon opening
- By default save Maestro Projects in version 5 format with an option to save in version 4 for backward compatibility
Support added for two new CIF file formats: “PDBx/mmCIF (*.cif)” and “Small Molecule CIF (.cif)”
Opening Maestro locally from LiveDesign is now supported on macOS, Linux, and Windows
Revamped splash screens & iconography: Modern visuals for an updated look and feel

Workflows & Pipelining [KNIME Extensions]

LiveDesign Admin node can take user credentials from the LiveDesign Connection node enabling SSO configuration

Target Validation & Structure Enablement

Protein Preparation

Protein The Protein Preparation Workflow now considers Epik states of ligands during the hydrogen-bond network optimization stage by default

Protein X-Ray Refinement

New sf2map.py script quickly generates an aligned x-ray map, given an input structure and a cif file containing structure factors

IFD-MD

Updated IFD-MD for automatic sampling of histidine tautomer states (HID, HIE): Consider induced fit effects simultaneously to resolve receptor tautomeric states and predict receptor and ligand conformations

Binding Site & Structure Analysis

SiteMap

Automatically apply Combined Mode which breaks down sites larger than 800 Å3

Mixed Solvent MD (MxMD)

Improved cryptic pocket identification with new mixed solvent molecular dynamics (MxMD) interface including customizable visualization (Beta): Gain a clearer understanding of candidate binding pockets on the protein surface with a new interface to set up and analyze MxMD simulations

Hit Discovery

Active Learning Applications

Faster time-to-results in AL-Glide and Glide using ZeroMQ mode for machine learning evaluation stage and Glide docking stage

Shape Screening

Additional similarity normalization schemes now available for Quick Shape and 1D Screening command. In addition to the max{O(A,A), O(B,B)} default can apply min{O(A,A), O(B,B)}, O(A,A), and O(B,B) where O(A,B) is the overlap between ligands A and B
Run Quick Shape and 1D Screens against a Phase pharmacophore hypothesis as the query
Improve speed of Quick Shape calculations with -limit and -NJOBS1D options that enable more efficient utilization of compute resources

Glide

Full release of Glide WS mode, previously known as WScore, to prioritize ligands for improved hit enrichment and pose prediction accuracy
- Leverages explicit water energetics to enhance the accuracy of protein-ligand poses and reduce experimentally inactive compounds in top-scoring virtual hits

Lead Optimization

FEP+

Gain deeper insights into receptor-ligand interactions with new Per-Residue Energy Decomposition in FEP Edge analysis
Perform categorical analysis in the Correlation Plot (FEP+) interface using classification matrices including common metrics such as Accuracy, Specificity, Recall, Precision, F1 Score, Cohen’s K and Kendall’s T
View reason why compounds are skipped in ABFEP calculations in the FEP+ Panel

Protein FEP+

Full release of Protein FEP+ Residue Scanning (with lambda dynamics)
- Workflow now generates an FMP archive to be loaded in the FEP+ Panel
- Web Services support

FEP+ Protocol Builder

Ability to run with either OPLS4 or OPLS5 force field
Added support for sampling of more residue protonation states
New option to skip active learning and perform exhaustive exploration of protocol parameter space

Quantum Mechanics

Return solvation entropy in implicit solvent calculations
distributed_frequencies.py workflow for numerical frequency calculations
Added support for isotope 11B in NMR calculations
Implicit solvent model SMD now has gradients and frequencies
E-sol now supports the CPCM-X implicit solvation method for rapid solvation energies (command line only)

Semi-Empirical Quantum Mechanics

Updated xTB to version 6.7.1 which uses the advanced solvent model CPCM-X

Macrocycles

Expanded list of predefined linkers from 5 to 22 for small molecule cyclization in macrocycle.py
Expanded list of side-chain bridges for peptide cyclization from 4 to 21 of the most commonly reported in the literature
Control spacers in macrocyclize.py via a CSV file of SMILES strings
Improvements to macrocycle alignment reproducibility and performance in tug_align script and Ligand Alignment Panel
Macrocycle sampling script can optionally output only macrocycle conformers
Easily create cyclic peptides from sequence on the command line with peptide_cyclize.py script

De Novo Design

AutoDesigner – R-group Design

Boost exploration of similar ligands with new AutoDesigner Similarity feature that scores output ideas based on similarity to a user-provided set of compounds
Added exhaustive PathFinder enumeration of all routes of the starting ligand using all available building blocks for those routes
Added recursive trimming of the final set of outputs to generate additional outputs
Improved logging including an overview of the number of compounds generated at various stages of the workflow

AutoDesigner – Core Design

Improved logging including an overview of the number of compounds generated at various stages of the workflow

Biologics Drug Discovery

Predict protein properties with automated machine learning model building using protein descriptors: Leverage AutoQSAR analysis to train, validate, and apply AI/ML models for biologics properties prediction
Predefined selection sets for quick access to TCR regions like CDRs, alpha/beta chains, and more

Materials Science

GUI for Quantum ESPRESSO

Product: Quantum ESPRESSO (QE) Interface

Phonon-dependent dielectric properties reported in the Phonon DOS viewer
Workflow action menu (WAM) for NMR calculations
Support for phonon calculations with DFT-D3
Improved cell relaxation protocol
Schrödinger-compatible Quantum ESPRESSO releases available at Github
Support for distributed phonon calculations
Control over maximum number of retries after failure via config file (command line)
Initial parameters and constraints preserved in the QM Convergence Monitor

KMC Charge Mobility

Product: MS Mobility

Compute KMC Charge Mobility: Improved speed with robust QM convergence (command line)

Materials Informatics

Product: MS Informatics

Formulation ML: Increased number of available steps for hyperparameter tuning
Formulation ML: Option to replace hyperparameter tuning steps with training time
Formulation ML: Visualization of atomic contributions from the feature importance analysis
Machine Learning Property: Updates to existing models
Machine Learning Property: Prediction of melting point for molecular solids
Machine Learning Property: Prediction of non-aqueous solubility of molecules

Coarse-Grained (CG) Molecular Dynamics

Product: MS CG

Automated CG Mapping: Speed up for mapping large molecules
Automated CG Mapping: Particle types and the number of occurrences reported

Penetrant loading simulations

Product: Penetrant Loading (PL)

Penetrant Loading: Differentiation between pre-existing water and added water

Reactivity

Product: MS Reactivity

Nanoreactor: Option to set the width of the biasing potential
Reaction Workflow: Anharmonic zero point energy (ZPE) added to Project Table
Reaction Workflow: Support for enumeration on sites in rings
Reaction Workflow: Option to automate the swap fragment with enumeration
Reaction Workflow: Preview of reaction diagram at the setup

Microkinetics

Product: MS Microkinetics

Microkinetic Modeling: Support for multistage MKM analysis
Microkinetic Modeling: Support for zoom in on plots in the viewer panel
Microkinetic Modeling: Increased default value for maximum integration time step

MS Maestro Builders and Tools

Adsorption Enumeration: Support for selection of reactive atoms by atom indices
Disordered System: Improved UI with reconfigured options for tabs and dialogs
Disordered System: Support for keeping selected molecules rigid with tangled-chain option
Meta Workflows: Support for radial distribution function analysis
Nanoparticle: Option to include only molecules with center of mass inside the particle

Classical Mechanics

Barrier Potential for MD: Option to remove barrier from input structures
Droplet: Support for entering random seed in building a droplet
Droplet: Support for randomized initial velocities
Evaporation: Support for full control over which profiles to plot
Evaporation: Support for applying barrier potentials
Evaporation: Option to set evaporation zone based on the distance from COM of the substrate
Evaporation: Improved loading speed for large input structures
MD Multistage: Improved relaxation protocol for ladder polymers
Refined default timestep for DPD particles
Thermostat and barostat settings set automatically for atomistic and coarse-grained systems
Stress Strain: Option to plot normal average stress
Thermophysical Properties: Option to return *.ene files (command line)
Trajectory Density Analysis: Output *.csv files

Quantum Mechanics

Adsorption Energy: Option to select between kcal/mol and kJ/mol for energy units
Adsorption Energy: All output entries incorporated in Project Table as subgroups
Adsorption Energy: Robust detection algorithm for valid input adsorbates
Adsorption Energy: Support for loading options from a Quantum ESPRESSO config file
Optoelectronic Film Properties: Prediction of molecular refractive indices
Optoelectronic Film Properties: Prediction of intersystem and reverse intersystem crossing rates
Optoelectronic Film Properties: Improved loading protocols for large input structures

Education Content

Life Science

New Tutorial: Protein pKa Prediction with Constant pH Molecular Dynamics
Updated Tutorial: Glide WS Evaluation of HSP90 Ligands

Materials Science

New Tutorial: Singlet-Triplet Intersystem Crossing Rate
New Tutorial: Modeling the Formation and Decomposition of Nitrosamines
New Tutorial: Atomic Layer Deposition
New Tutorial: Elemental Enumeration
New Quick Reference Sheet: Refractive Index
Updated Tutorial: Introduction to Multistage Quantum Mechanical Workflows

Docs Content

New documentation page to explore solutions for materials science applications and to identify the best fit for users’ interest
Panel images shown in the help topic of each panel

LiveDesign

What’s new in 2024-4

Biologics SAR Visualization: View, highlight, and analyze properties alongside sequence differences in the Sequence Viewer tool
Ligand Designer
- Upload multiple overlays in the Ligand Designer Configurations via the admin panel, and enable or disable the overlays during design sessions
- Rename ligands after editing and using the “predict pose” functionality to track and manage iterative design changes
R-group enumeration: Filter output products by computed properties
LiveDesign Learning: View the LiveDesign Learning Dashboard within Landing Pages
- *LiveDesign Learning is now called LiveDesign ML
LiveReport Management
- View row, column, and cell count for LiveReports in the LiveReport Picker
- Set up a LiveReport as Read-Only or Hidden during creation in an updated Create LiveReport dialog
- The following menu items have been consolidated into an “Edit LiveReport…” menu: Rename, Move to Folder, Make Read Only and Make Editable, Make Hidden and Make Visible
- Admins can update Read-Only LiveReports to make them editable
- Unhide a subset of compounds by clicking on a link in the LiveReport footer and selecting the compound IDs in a dialog
Documentation
- The ? button in LiveDesign now directs to online documentation as is done elsewhere in the Schrödinger platform. Users will be directed to log into their Schrödinger web account that will be verified with a code to their email; if they do not have a web account, they will need to sign up for one.
Copy values from the Form spreadsheet, table, and ID widgets to the clipboard
Search and filter for a column name in the Create MPO dialog when adding a constituent column from the LiveReport
Click a link in a 3D cell for a 3D Generic Entity and Explicit Ligand Designer to view the 3D structure in Maestro
Entities appear in a LiveReport more quickly after a reaction enumeration, R-group enumeration, file upload, Maestro upload, and LDClient upload
LDClient’s API for retrieving FFC columns “get_freefrom_column_by_id” will now include an “updated_at” field containing a long corresponding to the timestamp that the FFC was last updated
Landing Page: View recent published experimental data from the Compound’s Page

What’s Been Fixed

Adding an entity to an Advanced Search, by searching for its ID, would show an unresponsive dialog, and now shows a dialog that accept an ID.
Complex filters would not accept a pasted biologic sequence, and would fail to filter to that sequence, but now accept pasted biologic sequences.
Creating a LiveReport from a template in the Landing Page would not open the newly created LiveReport, and now opens the LiveReport in a new browser tab.
If an entity is imported to multiple projects, purging the entity from one of the projects will make it disappear from that project only. Only one file entry now appears in the Manage Files Dialog for a multi-entity import through CSV, ZIP, and FASTA.
String type of entity relationship metadata is supported in DI by adding a DI mapping for relationship and relationship metadata.
The column “Structure Class” has been renamed to “Entity Type” in the Data & Columns tree
Pasting aromatic structures into the sketcher would flip chirality on structures containing a pyrrole, and now the chirality is maintained.
When entering FFC date values via LDClient, users must now use the YYYY-MM-DD format. If the date is entered in an incorrect format, the system will return a 400 error with the message: ‘Date must be in YYYY-MM-DD format.’
Forms kanban widgets would show a pin icon beside the kanban tile, and now pin icon does not appear in kanban widgets.
Unpinning a row now works as expected.
Model results with multiple values in a cell would show a different order of values if the LiveReport was duplicated, and now show the same order in the duplicated LiveReport as the origin LiveReport.
The Guanidine group in Arginine incorrectly displayed a carbon with five bonds in the 3D visualizer, and now accurately represents the chemistry, showing the correct bonding structure for Arginine residues.
Reordering of R-group scaffolds will work for newly added or deleted scaffold without page refresh in SAR analysis(R-group decomposition).
The deleted docked poses will no longer reappear, ensuring a clean and organized workspace. Each new pose generated after clicking “predict pose” will be sequentially named, allowing for easy tracking (e.g., “docked_ligand_4” following the deletion of “docked_ligand_3”).
Changing the model’s name via Admin panel will get correctly reflected in the Ligand Designer.
Targets would remain visible in the 3D visualizer after unchecking the display checkbox and then selecting a different entity, and now the Target display checkbox remains unchecked and the Target is not visible.
Copying a LiveReport from one project to another would grant the destination project ACLs to any model within the LiveReport, even if the Model’s Protocol did not provide access to the destination project. Now, the model column will be “dummified” and not visible in the destination project if the Protocol does not provide access to that project.
Models with dependent parameterized models will not show an error when archived, and all of its dependent parameterized models will also get archived. A dialog box will appear stating the number of dependent parameterized models that will be archived.

Training & Resources

Online Certification Courses

Level up your skill set with hands-on, online molecular modeling courses. These self-paced courses cover a range of scientific topics and include access to Schrödinger software and support.

Learn more

Tutorials

Learn how to deploy the technology and best practices of Schrödinger software for your project success. Find training resources, tutorials, quick start guides, videos, and more.

Learn more

Other Resources

AI in Drug Discovery USA 2024

Posted on October 2, 2024October 2, 2024 by Anonymous

Conference

AI in Drug Discovery USA

Date & Time

October 21st-22nd, 2024

Location

Boston, Massachusetts

Schrödinger is excited to be participating in the AI in Drug Discovery USA conference taking place on October 21st – 22nd in Boston, Massachusetts. Join us for a presentation by Karl Leswing, Executive Director, Machine Learning at Schrödinger, titled “Latest advancements in machine learning-enhanced in silico design: Impact on a pipeline of drug discovery programs.”

Speaker:

Karl Leswing, Executive Director, Machine Learning, Schrödinger

Key Learning Objectives:

Using active learning with FEP+ for large-scale in silico fragment screens in hit discovery
Applying de novo design workflows for intelligent molecular core design
Leveraging experimental data for enhancing ADMET profiles in lead optimization using an interactive ML dashboard

Karl Leswing

Executive Director, Machine Learning, Schrödinger

Karl Leswing is the Executive Director for Machine Learning at Schrödinger. In this role he oversees the research and execution of machine learning applications for Schrödinger’s digital chemistry platform. In 2017 he was a visiting researcher at the Pande Lab working on using deep learning techniques for drug discovery. During that time he co-authored MoleculeNet, a benchmarking paper analyzing machine learning techniques for chemoinformatics. Karl received his undergraduate degree from the University of Virginia, and a Master’s in machine learning from Georgia Tech.

AAPS 2024 PharmSci 360

Posted on September 24, 2024October 1, 2024 by Anonymous

Conference

AAPS 2024 PharmSci 360

Date & Time

October 20th-23rd, 2024

Location

Salt Lake City, Utah

Schrödinger is excited to be participating in the AAPS 2024 PharmSci 360 conference taking place on October 20th – 23rd in Salt Lake City, Utah. Join us for presentations by Schrödinger scientists. Stop by booth #2503 to speak with us.

Presentations

OCT 21 | 10:00AM – 10:30AM

251 F Salt Palace Convention Center

Coarse-Grained Modeling of Nucleic Acid-Loaded Lipid Nanoparticle Formulations

Speaker:
Doug Grzetic, Senior Scientist I, Schrödinger

Abstract:
We will start off with a problem statement describing how the complicated nature of lipid nanoparticle formulations makes efficient formulation optimization a challenge. Additionally, the effectiveness of LNP formulations is believed to be strongly correlated to the LNP morphology, but this is difficult to characterize, making predictive, in silico measurements extremely valuable. Then we will provide a brief description of molecular modeling, emphasizing that for LNP self-assembly length-scales (~100 nm) coarse-grained modeling is required. In addition, high-throughput screening studies require that the building of CG models be automated as much as possible. We briefly review techniques for the automation of this process. We demonstrate the application of coarse-grained modeling to RNA-encapsulating LNPs, with a case study focusing on the Pfizer-BioNTech COVID-19 vaccine formulation.

OCT 21 | 3:15PM – 3:30PM

255 EF Salt Palace Convention Center

Modernize your arsenal of formulation tools with physics-based molecular simulation

Speaker:
Ben Coscia, Principal Scientist I, Schrödinger

Abstract:
The impact of physics-based molecular modeling and simulation on formulation is expanding rapidly with advancement of computer hardware and software algorithms. Cloud-based solutions enable individuals to access the world’s most powerful processors with just an internet connection. Machine learning algorithms continue to be leveraged towards improving the accuracy of our models and to guide high throughput simulation studies towards targeted properties. Despite this progress, one can argue that physics-based simulation is an underutilized technique, in large part due to slow adoption by non-experts. The purpose of this talk is to inform our audience of the accessibility of simulation and empower them to take the first steps towards interrogating their research questions with simulation, with specific emphasis on solubilization. We do this by example, describing two case studies which apply physics-based molecular simulation to gain insight into two different approaches that have direct implications on solubilizing poorly soluble APIs.

2024 AIChE Annual Meeting

Posted on September 24, 2024September 30, 2024 by Anonymous

Conference

2024 AIChE Annual Meeting

Date & Time

October 27th-31st, 2024

Location

San Diego, California

Schrödinger is excited to be participating in the 2024 AIChE Annual Meeting taking place on October 27th – 31st in San Diego, California. Join us for presentations by Schrödinger scientists. Stop by booth #521 to speak with us.

Presentations

OCT 28 | 8:00AM – 8:30AM

Hilton San Diego Bayfront Hotel, Sapphire Ballroom E

Accelerating Polymer Design with Targeted Properties Using Machine Learning and Physics-Based Models

Speaker:
Alex Chew, Principal Scientist I, Schrödinger

Abstract:
Designing new, industrially relevant polymers is challenging because of the need to optimize multiple materials’ properties simultaneously, which is expensive and often infeasible using traditional trial-and-error approaches. One possible solution to identifying promising polymeric materials is to employ a combination of machine learning and physics-based tools to screen the polymer design space and provide suggestions for new polymers that meet the criteria for an industrial application. In this work, we demonstrate a workflow that utilizes machine learning and molecular modeling approaches to design new polymers (specifically, polycarbonates) that satisfy five polymer properties, including the glass transition temperature, optical properties, and mechanical properties. Using a relatively modest dataset of fewer than 200 points, we developed quantitative structure-property relationship (QSPR) models to accurately predict the experimental polymer properties given the homo- or co-polymer structures and composition as input. Leveraging these computationally efficient QSPR models, we then screened over ~10,000 polymer structures that were generated through R-group enumeration tools. We used these QSPR predictions to create multi-parameter optimization scores to help down-select the large polymer space to ~10 promising candidates. We validated the predicted properties of the top polymer candidates using classical molecular dynamics simulations and density functional theory, which revealed reliable correlation between physics-based and QSPR approaches. Finally, we validated the computational predictions against experiments, which showed good agreement with QSPR and physics-based models. Our workflow demonstrates the usefulness of combining data-driven and physics-based approaches in designing new polymers given a small dataset, which is broadly useful for scientists interested in leveraging computer-aided strategies to innovate new materials while mitigating the need for extensive trial-and-error experimentation.

OCT 29 | 4:00PM – 4:30PM

Hilton San Diego Bayfront Hotel, Aqua 300 (AB)

Capturing the unmeasurable: How atomistic simulations are bringing understanding to interfacial phenomena

Speaker:
Andrea Browning, Director, Schrödinger

Abstract:
Most materials development at some point must consider an interface. Between adhesives and component parts, between matrix and filler in composite materials, and between atomic layers during assembly, the interface impacts the overall performance of the product. But these surfaces can be difficult to probe experimentally. Atomistic level modeling and simulation techniques such as quantum mechanics and molecular dynamics allows a window into the specific interactions that accumulate into the observed interfacial behavior. As simulation techniques and compute power has grown, we are now better able to explore how interfaces behave. However, there are still challenges remaining such as accounting for reactions at complex, multicomponent interfaces. From battery solid electrolyte interphases to dissolving polymers at tablet interfaces, simulations must capture discrete interactions that are important to the interface in order to be useful. This talk will review examples from various industries in how simulation of interfaces has developed and the role of technology exploration in Dr. Browning’s career evolution.

Computational Medicinal Chemistry School

Posted on September 18, 2024September 30, 2024 by Anonymous

Conference

Computational Medicinal Chemistry School

Date & Time

October 28th-30th, 2024

Location

Cambridge, Massachusetts

Schrödinger is excited to be a Founding Sponsor at the Computational Medicinal Chemistry School conference taking place on October 28th – 30th in Cambridge, Massachusetts. Join us for a presentation by Andreas Verras, Director at Schrödinger, titled “Beyond Potency: How Modeling can contribute to ADMET with structure based, ligand property, and machine learning approaches.”

Presentation

OCT 28 | 2:15 – 3:00 PM

Beyond Potency: How Modeling can contribute to ADMET with structure based, ligand property, and machine learning approaches.

Speaker:
Andreas Verras, Director, Schrödinger

Abstract:
Potency optimization is a general first step in drug discovery, but can be quickly overshadowed by other problems that make in vivo studies impossible. Optimizing absorption, metabolism, efflux and off-target toxicities is necessary to generate molecules that can interrogate your mechanism in an animal and ultimately go to the clinic. I will explore modeling approaches to understanding pharmacokinetic data; improving efflux, absorption, and eflux; and modeling some off target data. A combination of structure based, ligand based, and machine learning approaches is presented with some opinion on which problems they are most suited.

SEPAWA 2024

Posted on September 4, 2024September 30, 2024 by Anonymous

Conference

SEPAWA 2024

Date & Time

October 16th-18th, 2024

Location

Berlin, Germany

Schrödinger is excited to be participating in the SEPAWA 2024 conference taking place on October 16th – 18th in Berlin, Germany. Join us for a presentation by Jeff Sanders, Senior Principal Scientist at Schrödinger, titled “Beyond AI: The importance of Physics-based Modeling and Machine Learning to Develop New Cosmetic Products.”

Presentation

OCT 18 | 10:45 – 11:15

Room 15

Beyond AI: The importance of Physics-based Modeling and Machine Learning to Develop New Cosmetic Products

Speaker:
Jeff Sanders, Senior Principal Scientist

Abstract:
Despite the increasing popularity of machine learning and AI tools in consumer goods industries, methods are often applied ad hoc, lacking systematic data sampling or a comprehensive understanding of resulting ML models. This can lead to a knowledge gap between ML/AI proponents and researchers, especially when datasets are small or not easily transferable. Newer methodologies, like active learning, aim to bridge this gap by integrating physics-based simulation and experimental data to not only construct models but also identify blind spots in relevant property space coverage. In cosmetic formulation research projects where chemistry and composition are crucial, active learning can be utilized to develop models and guide experimentation in an iterative feedback loop, offering unique insights into model performance. By prioritizing chemistry first, many physicochemical descriptors can be generated using physics-based simulations, thereby enriching experimental datasets and averting the “blackboxing” of valuable models.

The importance of human know-how in AI execution for R&D

Posted on August 28, 2024June 2, 2025 by Anonymous

The importance of human know-how in AI execution for R&D

How Schrödinger’s materials science domain experts ensure partner success

Connect with us

The importance of human know-how in AI execution for R&D

Overview

As artificial intelligence (AI) and machine learning (ML) technologies rapidly advance, materials scientists, executives, and R&D professionals are being tasked with developing an AI/ML strategy to drive innovation. Yet while AI/ML tools may advertise the appeal of push-button innovation, this is almost never the reality. Even the term AI itself can be a misleading buzzword.

Schrödinger is uniquely positioned to partner with materials R&D teams to execute AI/ML strategies that deliver true business value because we leverage the proven accuracy of physics-based modeling, the speed and scale of machine learning, and the deep domain expertise of our materials scientists. So while AI/ML technologies can be transformative, gaining a competitive advantage with AI is only possible with the talent, vision, and know-how of people who can utilize and direct these technologies towards meaningful, impactful outcomes.

Personalized human support is critical for project efficiency and productivity

Schrödinger’s professional software support is unique in providing domain expertise, personalized assistance, and reliable service. This human difference ensures that users can fully leverage the potential of digital tools to address complex issues through tailored guidance and support.

Schrödinger Support	Benefits
24-7 global support from a team of experts	Faster project timelines
On-site and virtual assistance with expert application scientists	Technical discussions, troubleshooting, and knowledge transfer
Extensive online training, tutorials, and resources	Skilled users to maximize outcomes
Expert-led customized modeling service packages	Novel state-of-the-art research, knowledge transfer and improved success rates

“As a former industrial modeler at Boeing, I understand that the needs of each client are unique. That’s why we focus on delivering personalized support that addresses their specific challenges in R&D digitization.”

Andrea Browning Director of Polymers and Soft Matter

Schrödinger technical experts increase capacity and capability

Through strategic partnerships or customized contract research, Schrödinger’s team of expert scientists work closely with customers to tackle challenging problems by deploying digital chemistry strategies to guide rapid materials design and optimization. By working as a team, we share the same challenges and goals. These expert-driven collaborations transform challenges into opportunities, driving innovation and delivering exceptional results.

“Over the course of 20 years working with companies in this sector, I am convinced that collaboration is the key to delivering value, by listening to industry needs and tailoring the modeling solution.”

Simon Elliott Director of Atomic Level Process Simulation

Expertise and industry-grade software can be the difference between project success and failure

It is common for materials science researchers to piece together a variety of modeling and simulation tools. Free software packages are often limited by outdated documentation, insufficient support, and obstacles to integration and automation. The human element is one of the key differentiators between industry-grade software like Schrödinger and less sophisticated software packages.

Schrödinger has a large team working on user experience, technical support, and education. This team ensures that clients can seamlessly access the science underpinning the simulations and use it at a scale that allows them to get their job done.

“Having used free software extensively in the past, I now realize how much time I was spending on troubleshooting rather than actual research. Schrödinger’s solutions are a game-changer for productivity.”

Pavel Dub Senior Principal Scientist and Product Manager

Advanced physics-based modeling and AI/ML software for R&D teams

Schrödinger brings more than 30 years of scientific innovation and deep domain expertise to empower R&D teams to solve their unique challenges with computational approaches.

View the flyer

Recent Testimonials

“By working closely with Schrödinger experts, we were impressed by how fast we were able to learn to apply molecular simulations, even with no prior modeling experience. Our collaborations have been very successful, not only because of our satisfaction with Schrödinger’s advanced technologies, but also because of their level of scientific expertise, support, and collaborative openness.”
Martin SettleSenior Research Manager, Reckitt

The resin design and incubation team at SABIC worked closely with Schrödinger’s material science team to build accurate machine learning (ML) models to speed up the discovery of new polymers. “These computational results are highly promising and can potentially shorten our polymer innovation timelines from traditionally a couple of years to only a couple of months.”
Vaidya RamakrishnanStaff Scientist, SABIC

“Schrödinger provides us with more than just software as part of our service agreement—they are a true partner in our research. With an office here in Japan, Schrödinger scientists and engineers are easily accessible and able to collaborate in-person with our team.”
Nobuyuki N. MatsuzawaExecutive Engineer, Panasonic Industry Co., Ltd.

Software and services to meet your organizational needs

Software Platform

Deploy digital materials discovery workflows with a comprehensive and user-friendly platform grounded in physics-based molecular modeling, machine learning, and team collaboration.

Learn more

Research Services

Leverage Schrödinger’s expert computational scientists to assist at key stages in your materials discovery and development process.

view services

Support & Training

Access expert support, educational materials, and training resources designed for both novice and experienced users.

Explore Resources

Accelerated in silico discovery of SGR-1505: A potent MALT1 allosteric inhibitor for the treatment of mature B-cell malignancies

Posted on August 26, 2024March 13, 2025 by Anonymous

SEP 12, 2024

Accelerated in silico discovery of SGR-1505: A potent MALT1 allosteric inhibitor for the treatment of mature B-cell malignancies

Abstract:

MALT1 (Mucosa-associated lymphoid tissue lymphoma translocation protein 1) is a component of the MALT1-BCL10-CARD11 complex downstream from the Bruton Tyrosine Kinase (BTK) on the B-cell receptor signaling pathway. MALT1 is a key mediator of nuclear factor kappa B (NF-κB) signaling, which is the main driver of a subset of B-cell lymphomas. MALT1 is considered a potential therapeutic target for several subtypes of non-Hodgkin’s B-cell lymphomas and chronic lymphocytic leukemia (CLL), including tumors with acquired BTK inhibitor (BTKi) resistance. Constitutive activation of the NF-κB is a molecular hallmark of activated B cell-like diffuse large B cell lymphoma (ABC-DLBCL), and MALT1 may have utility as a treatment option for ABC-DLBCL. Furthermore, a third-party MALT1 inhibitor recently showed strong anti-tumor activity in mature B cell malignancies from Phase 1 studies.

By applying advanced physics-based modeling techniques, including combining free energy calculations with machine learning methods and chemistry-aware compound enumeration workflow, the Schrödinger team explored extensive sets of de novo design ideas to quickly identify a novel hit series with an in vivo tool molecule to establish an in vivo PD and efficacy mouse model early on in the project. Multi-parameter optimization (MPO) allowed efficient prioritization of molecules with good potency and drug-like properties during lead optimization. This led to the discovery of a highly potent MALT1 inhibitor, SGR-1505, with a well-balanced property profile in under a year, with only 78 compounds synthesized in the lead series and 129 compounds overall. SGR-1505 is a potent and orally available allosteric MALT1 inhibitor. It demonstrated strong anti-tumor activity alone and in combination with BTK inhibitors in multiple in vivo B-cell lymphoma xenograft models. Currently, a Phase 1 clinical trial with SGR-1505 in patients with mature B-cell neoplasms is ongoing (NCT05544019).

Webinar Highlights:

Discover how free energy calculations, amplified by machine learning methods, led to the discovery of a highly potent MALT1 inhibitor, SGR-1505
Learn how the team used an MPO scoring function consisting of FEP+-based predictions of affinity and solubility, physics-based predictions of permeability, and predictions of lipophilicity to optimize compounds
Ask questions to gain further insight from the speakers to apply to your work

Our Speakers

Goran Krilov

Senior Director, Schrödinger

Dr. Goran Krilov is a senior Director of Computational Chemistry at Schrödinger’s Therapeutic Group. For the past twenty five years, his work has focused on developing and applying cutting-edge computational chemistry techniques to problems in biophysics and drug discovery. He has led the modeling efforts on a number of internal projects as well as external collaborations in oncology and neurogenerative diseases, resulting in two clinical candidates currently undergoing Phase I trials. Prior to joining Schrödinger, Dr. Krilov has worked in both industry and academia, including IBM, Boston College snd Strand Life Sciences.

Zhe Nie

Executive Director, Schrödinger

Dr. Zhe Nie is the Executive Director of Medicinal Chemistry at Schrödinger’s Therapeutic Group. She has been leading multiple wholly owned and partnered drug discovery programs at Schrödinger. Most recently, she led Schrödinger’s MALT1 discovery project team, successfully developed the small molecule drug SGR-1505 (Schrödinger’s first internal clinical asset currently in Ph1) applying Schrödinger’s computational platform. It took less than two years from the start of the project to the selection of the clinical candidate. She also led the DLK collaboration project with Takeda Pharmaceuticals which discovered a potent, selective, and brain-penetrate DLK inhibitor as a promising preclinical candidate for the treatment of neurodegenerative diseases using Schrödinger’s computational platform. She has extensive experiences in applying advanced computational tools to assist in the design of small molecule drug candidates. She previously worked at Takeda, Celgene and Quanticel Pharmaceuticals (acquired by Celgene), led and contributed to advancing multiple small molecule drugs to the clinics including TAK-960, TAK-659 and CC-90011.

Webinar Series: From Molecules to Materials Applications

Posted on August 16, 2024January 22, 2025 by Anonymous

Webinar Series

From Molecules to Materials Applications

Date & Time

September 11th – October 8th, 2024
18:00 IST

Location

Virtual

Molecular modeling is a powerful computational technique widely used in materials science to predict and understand the properties and behavior of materials at the molecular level. By simulating the interactions between atoms and molecules, researchers can explore the structural, mechanical, electronic, and thermal properties of various materials and gain a deeper understanding. Molecular modeling encompasses a range of methods, including molecular dynamics, quantum mechanics, and coarse grained simulations, each providing unique insights into material properties and guiding experimental efforts.

The integration of molecular modeling into materials science can accelerate the development of advanced materials for applications in pharmaceuticals, Fast Moving Consumer Goods, electronics, energy storage, catalysis, and more.

This webinar series “From molecules to Materials Applications” will delve into molecular modeling techniques and their transformative impact on Materials Science research using the Schrödinger Materials Science tools.

Agenda

SEPT 11 | 18:00 IST

Molecular Modeling: A Key to Solving Real-Life Challenges in Pharma Formulations

Speaker:
Sudharsan Pandiyan, Principal Scientist II, Schrödinger

Abstract:
The demand for innovative drug delivery methods has driven researchers to explore the intricate structure-property relationships within pharmaceutical formulations. Quantum Mechanical (QM) and Molecular Dynamics (MD) simulations are powerful tools for understanding these formulations at a molecular level. Key areas of interest in pharmaceutical sciences include chemical stability, reactivity, molecular degradation, impurity profiling, excipient selection, and polymorph prediction. A thorough understanding of the Active Pharmaceutical Ingredient (API) is essential before embarking on the formulation development process. The Schrödinger Materials Science Suite (MS-Suite) offers comprehensive computational workflows to predict spectra (IR, Raman, NMR, UV-Visible, XRD) and assess the API’s behavior under varying pH conditions, including its degradation pathways and chemical reactivity. Recent advancements in GPU technology have significantly accelerated MD simulations, enabling previously unattainable time scales. This dramatic speedup, combined with predictive accuracy, is poised to revolutionize the use of MD simulations in pharmaceutical formulation development. MD-based workflows can help us address critical formulation design questions on physical stability of formulations, phase transitions, miscibility, solubility and diffusion of API through membranes, morphology, excipient compatibility, encapsulation, and coating selections. This presentation will highlight several successful case studies that demonstrate these capabilities from a molecular perspective.

SEPT 18 | 18:00 IST

Harnessing Molecular Modeling to transform innovation in Polymeric Materials and Consumer Packaged Goods

Speaker:
Sriram Krishnamurthy, Senior Scientist I, Schrödinger

Abstract:
Polymeric materials and consumer packaged goods (CPGs) hold significant importance in both industrial and everyday contexts, impacting numerous aspects of modern life. Polymers play a crucial role in various industries due to their versatility and wide range of applications like construction, electronics, healthcare, and automotive. As scientific research and innovation continue to advance, polymeric materials remain at the forefront of development. In the context of consumer packaged goods (CPG), which significantly influence our daily lives, there is an ongoing push to create products that meet evolving consumer demands, such as healthier food options and more sustainable packaging solutions. Polymeric materials and CPGs are deeply interconnected in their importance to modern society. Molecular modeling has emerged as a transformative tool in the design and optimization of these materials. By providing deep insights into molecular interactions and material properties, molecular modeling accelerates the development of novel, efficient, and sustainable materials. This approach not only enhances our understanding of material behavior, but also facilitates the innovation of advanced solutions tailored to the specific needs of modern consumer products. This webinar will highlight Schrödinger’s Materials Science tools that can accelerate R&D efforts in these scientific domains. We will showcase practical case studies to tackle key problems and identify areas where molecular modeling can be applied.

SEPT 25 | 18:00 IST

Efficient Computation of Process Parameters for Controlling the Chemistry of Deposition or Etch

Speaker:
Simon Elliott, Research Leader, Schrödinger

Abstract:
We present a variety of computational techniques for understanding, controlling and improving deposition and etch processes. The emphasis is on choosing the right technique for the research question and time available. The same computational techniques can be used to investigate other gas-surface processes, such as catalysis or sensing. Different chemical processes can be in competition when a solid surface is treated with a gaseous reagent and the outcome is determined by conditions such as temperature and pressure. For instance, continuous deposition (CVD) may take over from self-limiting deposition (ALD) as the temperature is raised. Or temperature may dictate which material is deposited; in the case presented here, ruthenium oxide film is deposited from RuO4+H2 in experiments at 75°C, whereas Ru metal is obtained at 100°C and above. Ru is being investigated as an electroplating seed layer in electronics, as a capacitor electrode and as a heterogeneous catalyst – all applications that require metal rather than oxide. We show that thermodynamics based on density functional theory (DFT) is a computationally-efficient approach for distinguishing between the possible surface-gas processes. The temperatures and pressures for crossover between different chemistries can be estimated, with the accuracy depending on how entropy, coverage and diffusion are treated. We use DFT to examine the conditions of stability for Ru metal, hydride, hydroxide and oxide with respect to H2 and RuO4 reagents, and so explain the crossover from oxide to metal film just below 100°C. We point out how to balance the cost (in terms of researcher time and computer time) against the benefit that each level of accuracy can offer. In the second part of the talk, we introduce Microkinetic Modelling, a new Schrödinger capability for examining the overall kinetics of gas-surface chemistry by solving the coupled kinetic rate equations of its constituent elementary reaction steps. This allows the simulation of macroscopic parameters such as sticking coefficients that can be experimentally measured and used as inputs for fluid dynamics simulations. We first outline the computational scheme, where elementary steps and their activation free energies have been computed with DFT. The resulting microkinetic model for alumina ALD yields measurable quantities (e.g. growth rate) as a function of temperature and pressure, which are validated against experiment. Variation with pressure can account for penetration depth and conformality within high aspect ratio features. The two cases discussed in this talk thus illustrate how atomic-scale DFT can be embedded into higher-level computational schemes for accurate and achievable prediction of the conditions and parameters for controlling chemical processes.

OCT 1 | 18:00 IST

How Physics-based Modeling and Machine Learning Enable Accelerated Development of Battery Materials

Speaker:
Garvit Agarwal, Senior Scientist II, Schrödinger

Abstract:
The rapid advancements in rechargeable Li-ion battery (LIB) technology over the last decade has revolutionized several key industries such as transportation and consumer electronics. However, new battery chemistries are needed to meet the rapidly growing demand and to improve the power density, safety, reliability, and lifetime of LIBs. Molecular modeling has become an integral part of the design cycle of new battery chemistries. Accurate physics-based modeling enables rapid evaluation and screening of large chemical and material design space thereby, helping industries reduce the time required to bring the new technology to the market. In this webinar, we will introduce the latest technological innovations in Schrödinger’s digital chemistry platform for battery materials design. In particular, the webinar will focus on examples to demonstrate the application of automated solutions for accurate prediction of thermodynamic stability and voltage profile of cathode materials, ion diffusion pathways and kinetics in electrode materials, transport properties of liquid electrolytes and modeling the nucleation and growth of solid electrolyte interphase (SEI) layers using Schrödinger’s SEI simulator module. We will also introduce an automated generalized framework for the development of customized machine learning force fields for complex materials such as liquid electrolytes, inorganic cathode coatings and solid polymer electrolytes, paving the way for efficient design of novel materials for next generation batteries.

OCT 8 | 18:00 IST

Accelerating the Design of Asymmetric Catalysts with Schrödinger’s Digital Chemistry Platform

Speaker:
Saientan Bag, Senior Scientist I, Schrödinger

Abstract:
Asymmetric catalysis has become an integral part of the science-driven technological revolution in the second half of the 21st century, leading to decreased energy demands, sustainable chemical processes and the realization of “impossible” transformations. Asymmetric catalysis based on chiral transition-metal complexes plays an important role in the synthesis of single-enantiomer drugs, perfumes and agrochemicals. The importance of the field is recognized by two Nobel Prize Awards in 2001 (transition-metal catalysis) and 2021 (organocatalysis). Asymmetric catalysts are traditionally designed by experimental trial-and-error methods, which are resource-, time- and labor-consuming, and thus extremely expensive. Digital methods offer the opportunity to expedite catalyst design. Until recently, computational chemistry, typically quantum chemical studies, indirectly contributed to asymmetric catalyst design by providing rationalization for the mechanism of generation of chirality. With the development of more advanced methods, algorithms and an included layer of automation, computational catalysis is now providing the possibility for direct asymmetric catalyst design. In this webinar, I will demonstrate how Schrödinger’s advanced digital chemistry platform can be used to accelerate the direct design and discovery of asymmetric catalysts.

AI/ML meets physics-based simulations: A new era in complex materials design

Posted on August 14, 2024November 6, 2025 by Anonymous

SEP 27, 2024

AI/ML meets physics-based simulations: A new era in complex materials design

The simulation of materials properties using physics-based approaches, such as density functional theory (DFT) and molecular dynamics (MD), has long been successful in providing insights into structure-property relationships and subsequently aiding in the design of novel materials. In recent years, AI/machine learning (ML) has been used extensively in conjunction with physics-based modeling techniques to greatly accelerate materials innovation. The accuracy and generalizability of physics-based modeling improves the performance of AI/ML models and enables them to be used effectively even in small-data regimes. Conversely, the speed and flexibility of AI/ML help bridge the time- and spatial- scale limitations of physics-based models, creating a synergistic approach that optimizes both predictive accuracy and computational efficiency.

In this webinar, we demonstrate the application of this combined approach in designing materials and formulations across diverse materials science applications, from battery electrolytes and fuel mixtures to thermoplastics and OLED devices.

Key Learning Objectives:

Understand how DFT descriptors enhance the accuracy of AI/ML models for optoelectronic molecules and battery electrolytes
Discover how MD simulation descriptors improve AI/ML models for the viscosity of organic molecules
Explore the use of Schrödinger’s automated Formulation Machine Learning solution to:
- Train AI/ML models for the solubility of APIs in binary solvents
- Predict the motor octane number of hydrocarbons
Learn about advances in AI/ML force field technology (QRNN) and its application in modeling the bulk properties of inorganic cathode coating materials

Our Speaker

Anand Chandrasekaran

Senior Principal Scientist, Schrödinger

Anand Chandrasekaran joined Schrödinger in 2019 and he is currently the Product Manager of MS-Informatics. His expertise is in applying machine learning to different areas in Materials Science and computational modeling. He graduated from the group of Prof.Nicola Marzari in the Swiss Federal Institute of Technology, Lausanne with a PhD in Materials Science. Before joining Schrödinger, Anand also worked in the group of Prof. Rampi Ramprasad on a number of topics including polymer informatics, machine-learning force-fields, and machine-learning for electronic structure calculations.