MFMichael E. Fortunato

Michael E. Fortunato

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Michael E. Fortunato has a total of 24 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Applying Active Learning toward Building a Generalizable Model for Ni-Photoredox Cross-Electrophile Coupling of Aryl and Alkyl Bromides

    by Lucas W. Souza, Nathan D. Ricke, Braden C. Chaffin, Mike E. Fortunato, Shutian Jiang, Cihan Soylu, Thomas C. Caya, Sii Hong Lau, Katherine A. Wieser, Abigail G. Doyle, et al.

    When developing machine learning models for yield prediction, the two main challenges are effectively exploring condition space and substrate space. In this article, we disclose an approach for mapping the substrate space for Ni/photoredox-catalyzed cross-electrophile coupling of alkyl bromides and aryl bromides in a high-throughput experimentation (HTE) context. This model employs active learning (in particular, uncertainty querying) as a strategy to rapidly construct a yield model. Given the vastness of substrate space, we focused on an approach that builds an initial model and then uses a minimal data set to expand into new chemical spaces. In particular, we built a model for a virtual space of 22,240 compounds using less than 400 data points. We demonstrated that the model can be expanded to 33,312 compounds by adding information around 24 building blocks (<100 additional reactions). Comparing the active learning-based model to one constructed on randomly selected data showed that the active learning model was significantly better at predicting which reactions will be successful. A combination of density function theory (DFT) and difference Morgan fingerprints was employed to construct the random forest model. Feature importance analysis indicates that key DFT features that are related to the reaction mechanism (e.g., alkyl radical LUMO energy) were crucial for model performance and predictions on aryl bromides outside the training set. We anticipate that combining DFT featurization and uncertainty-based querying will help the synthetic organic community build predictive models in a data-efficient manner for other chemical reactions that feature large and diverse scopes.


  • Open Access

    Computer‐aided evaluation and exploration of chemical spaces constrained by reaction pathways

    by Itai Levin, Michael E. Fortunato, Kian L. Tan, Connor W. Coley

    The processes of molecular design and synthetic route selection are necessarily intertwined during discovery. Computational tools have been developed to facilitate synthesis planning, but in a discovery setting, finding a single route to a single molecule of interest may be less important than finding a route that enables rapid access to a library of analogs. Here, we demonstrate how we can estimate route “diversifiability” and use it as a criterion during route selection. We illustrate how the chemical space of synthetically accessible analogs is influenced by properties of alternative starting materials or constraints on their cost. Finally, we integrate these analyses with a synthesizability‐constrained hit expansion workflow in a virtual screening pipeline for focused library expansion around putative hits to support molecular optimization. As medicinal chemistry and adjacent fields shift toward more autonomous design and synthesis of new molecules, it will be increasingly important to embed considerations of synthesizability into molecular design to ensure that computational recommendations are actionable.


  • Open Access

    Introduction to In Silico Synthesis of Polymers via PySIMM [Article v1.0]

    by Alexander G. Demidov, B. Lakshitha A. Perera, Michael E. Fortunato, Sibo Lin, Coray Colina

    Pysimm is a framework for molecular simulations of polymers and polymer-based nanostructures, which enables their direct chemical synthesis and preparation. Pysimm facilitates the understanding of novel, amorphous, processable materials for a broad range of applications, including heterogeneous catalysts, adsorbents and gas storage materials, as well as protein-polymer conjugates. This tutorial provides a detailed guide on the construction of atomistic and united-atom models of polymers using Pysimm: an open-source Python Application Programming Interface for atomistic molecular simulations. The API complements and simplifies the work of widely known molecular simulation software, such as LAMMPS, CASSANDRA, NAMD and Amber. Readers should be familiar with the basic concepts of atomistic molecular simulations, as well as the basic knowledge of Python programming language, before attempting to follow this tutorial. This work is separated into 3 main sections. First, the process of building an atomic-level model of a polymer chain from its repetitive units is described. The second section shows how to work with existing forcefields, and how Pysimm can automatically read, recognize, and assign appropriate Force Field parameters to a molecule. The final section discusses how to use Pysimm to construct polymer chains with pre-specified tacticity. The section is also available in the form of an interactive Jupyter notebook tutorial outlining simple guidelines to construct polymer models.


  • Predicting In Vivo Compound Brain Penetration Using Multi-task Graph Neural Networks

    by Seid Hamzic, Richard Lewis, Sandrine Desrayaud, Cihan Soylu, Mike Fortunato, Grégori Gerebtzoff, Raquel Rodríguez-Pérez

    Assessing whether compounds penetrate the brain can become critical in drug discovery, either to prevent adverse events or to reach the biological target. Generally, pre-clinical in vivo studies measuring the ratio of brain and blood concentrations (Kp) are required to estimate the brain penetration potential of a new drug entity. In this work, we developed machine learning models to predict in vivo compound brain penetration (as LogKp) from chemical structure. Our results show the benefit of including in vitro experimental data as auxiliary tasks in multi-task graph neural network (MT-GNN) models. MT-GNNs outperformed single-task (ST) models solely trained on in vivo brain penetration data. The best-performing MT-GNN regression model achieved a coefficient of determination of 0.42 and a mean absolute error of 0.39 (2.5-fold) on a prospective validation set and outperformed all tested ST models. To facilitate decision-making, compounds were classified into brain-penetrant or non-penetrant, achieving a Matthew's correlation coefficient of 0.66. Taken together, our findings indicate that the inclusion of in vitro assay data as MT-GNN auxiliary tasks improves in vivo brain penetration predictions and prospective compound prioritization.


  • Data Augmentation and Pretraining for Template-Based Retrosynthetic Prediction in Computer-Aided Synthesis Planning

    by Michael E. Fortunato, Connor W. Coley, Brian C. Barnes, Klavs F. Jensen

    This work presents efforts to augment the performance of data-driven machine learning algorithms for reaction template recommendation used in computer-aided synthesis planning software. Often, machine learning models designed to perform the task of prioritizing reaction templates or molecular transformations are focused on reporting high-accuracy metrics for the one-to-one mapping of product molecules in reaction databases to the template extracted from the recorded reaction. The available templates that get selected for inclusion in these machine learning models have been previously limited to those that appear frequently in the reaction databases and exclude potentially useful transformations. By augmenting open-access data sets of organic reactions with explicitly calculated template applicability and pretraining a template-relevance neural network on this augmented applicability data set, we report an increase in the template applicability recall and an increase in the diversity of predicted precursors. The augmentation and pretraining effectively teaches the neural network an increased set of templates that could theoretically lead to successful reactions for a given target. Even on a small data set of well-curated reactions, the data augmentation and pretraining methods resulted in an increase in top-1 accuracy, especially for rare templates, indicating that these strategies can be very useful for small data sets.