skip to main content


Search for: All records

Creators/Authors contains: "Vakili, Pirooz"

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Major histocompatibility complex Class I (MHC-I) molecules bind to peptides derived from intracellular antigens and present them on the surface of cells, allowing the immune system (T cells) to detect them. Elucidating the process of this presentation is essential for regulation and potential manipulation of the cellular immune system. Predicting whether a given peptide binds to an MHC molecule is an important step in the above process and has motivated the introduction of many computational approaches to address this problem. NetMHCPan, a pan-specific model for predicting binding of peptides to any MHC molecule, is one of the most widely used methods which focuses on solving this binary classification problem using shallow neural networks. The recent successful results of Deep Learning (DL) methods, especially Natural Language Processing (NLP-based) pretrained models in various applications, including protein structure determination, motivated us to explore their use in this problem. Specifically, we consider the application of deep learning models pretrained on large datasets of protein sequences to predict MHC Class I-peptide binding. Using the standard performance metrics in this area, and the same training and test sets, we show that our models outperform NetMHCpan4.1, currently considered as the-state-of-the-art.

     
    more » « less
    Free, publicly-accessible full text available August 17, 2024
  2. Abstract

    Early phases of green material development can be accelerated by identifying driving factors that control material properties to understand potential tradeoffs. Full investigation of fabrication variables is often prohibitively expensive. We propose a pared‐down design of experiments (DOE) approach to identify driving variables in limited data scenarios using tunable polydimethylsiloxane (PDMS) foams made via sacrificial templating as an example system. This new approach systematically determines the dependencies of porosity, transparency, and fluid flow by varying the template particle size and packing while using a more sustainable solvent. Factor screening identified template particle size and packing density as the driving factors for foam performance by controlling pore size and interconnectivity. The framework developed provides a robust, foundational understanding of how to green and tune a novel material's properties using an efficient and effective exploration of the design space. Recommendations for applying this method to a broad suite of experiments are provided.

     
    more » « less
  3. null (Ed.)
  4. Many materials systems comprise complex structures where multiple materials are integrated to achieve a desired performance. Often in these systems, it is a combination of both the materials and their structure that dictate performance. Here the authors layout an integrated computational–statistical–experimental methodology for hierarchical materials systems that takes a holistic design approach to both the material and structure. The authors used computational modeling of the physical system combined with statistical design of experiments to explore an activated carbon adsorption bed. The large parameter space makes experimental optimization impractical. Instead, a computational–statistical approach is coupled with physical experiments to validate the optimization results. 
    more » « less
  5. By virtue of their extensive potential in energy conversion and storage, catalysis, photocatalysis, adsorption, separation and life science applications, significant interest has been devoted to the design and synthesis of hierarchical porous materials. The main factors which determines the performance of hierarchical porous materials for an application include structure (pore size, porosity, tortuosity), materials (scaffold, dopants) and operating conditions. Traditionally, these hierarchical porous materials are synthesised and fabricated through a manual trial and error procedure, which is an expensive and time-consuming approach. However, there have been significant advances in mathematical, computational and engineering tools toward solving and optimising multiscale descriptions of physical phenomena. This motivates a computational-aided framework to tailor the fabrication of hierarchical porous materials to be optimised in performance for their specific application. In this work, a reactive-transport system in porous media is modelled using computational fluid dynamics. While microscale descriptions are too computationally expensive and macroscale models fail to accurately describe a physical phenomena in specific parts of computational domains, hybrid - or multiscale - algorithms, are used. Using the information provided by the numerical simulation, multiscale model-based design of experiments are developed to optimise the material’s performance on their particular usage. It is proposed that hierarchical multiscale modeling offers a systematic framework for identification of the important scales and parameters where one should focus experimental efforts on. 
    more » « less