Skip to main content
  • Discoverability Visible
  • Join Policy Open/Anyone
  • Created 08 Sep 2021

Are you interested in learning how to use artificial intelligence for analysis of biological data? Experts from NCI, Frederick National Laboratory for Cancer Research, Lawrence Livermore National Laboratory, and NVIDIA will provide real-world examples on using AI for 1) drug development, 2) image analysis, 3) molecular data, and 4) multimodal data. The one-hour seminars will be offered monthly from May 25, 2021 through September 23, 2021 and are hosted jointly by the NCI Bioinformatics Training and Education Program (BTEP) and the NCI Data Science Learning Exchange. See the descriptions, below for more information or contact NCIDataScienceLearningExchange@mail.nih.gov.

AI in Drug Development, presented by the ATOM consortium

Date/Time: May 25th, 2021, 1 – 2 pm ET

Recording: https://btep.ccr.cancer.gov/wp-content/uploads/BTEP-AI-Seminar-Series-1-2021-05-25-13-02-08-1.mp4

Slides: here

Presenter: Jonathan Allen PhD, Computational Scientist, Lawrence Livermore National Laboratory

Title: Building data-driven small molecule property prediction models with AMPL

Description: This talk will introduce basic concepts in building small molecule property prediction using machine learning models trained on data collected from experimental assays. Practical challenges will be considered, starting with limitations in data collection and curation through to model selection for property prediction applications. The ATOM (Accelerating Therapeutics for Opportunities in Medicine) Modeling Pipeline (AMPL) will be used to provide concrete examples for building models and data visualization.

AI in Image Analysis, presented by CCR: AIR and High Throughput Imaging Facility

Date/Time: June 15, 2021, 11 am – 12 pm ET

Recording: https://btep.ccr.cancer.gov/wp-content/uploads/BTEP-AI-Seminar-Series-2-2021-06-15-11-00-57.mp4?=2

Slides: Gianluca’s slidesG Tom’s slides

Presenters: Gianluca Pegoraro PhD (Staff scientist) , G Tom Brown MD/PhD (Staff clinician), Center for Cancer Research, NCI

Title: Overview of Deep Learning Applications in Bioimaging and Digital Pathology

Description: Deep learning is a subclass of machine learning which has shown excellent performance in learning tasks on unstructured data, such as digital images. This talk consists of two parts. In the first part , we will discuss some recent development of machine learning, with emphasis on the cellular object segmentation and tracking, image classification, and image restoration. The second part focuses on applications in digital pathology where we will talk about data acquisition, curation, and cleaning, as well as approaches to deep-learning. By the end of this part, you should be aware of potential pitfalls than confound results and have a better understanding of what it takes to carry out a deep-learning project in digital pathology.

AI in Molecular Data, presented by NVIDIA

Date/Time: July 15, 2021, 1 – 2 pm ET

Presenter: Avantika Lal PhD, Senior Scientist, Deep Learning and Genomics, NVIDIA

Title: Machine Learning Tools to Analyze Gene Expression and Regulation

Description: This talk will describe machine learning and deep learning methods to analyze bulk and single-cell RNA sequencing data, as well as deep learning models that integrate epigenetic data to decipher the regulatory networks underlying gene expression.

Presentation: here

Recording: here

AI for Multimodal Data, presented by members of the Strategic and Data Science Initiatives, Frederick National Laboratory for Cancer Research.

Date/Time: Sep 23, 2021, 1 – 2 pm ET

Slides: here

Recording: here

Presenters: George Zaki, Bioinformatics Manager, Strategic and Data Science Initiatives (SDSI), Frederick National Laboratory for Cancer Research (FNL), Pinyi Lu, Bioinformatics analyst, SDSI, FNL

Title: Building Predictive Models From Multimodal Data Using Machine Learning

Description: In this talk, we will highlight two examples for building predictive models from multi modal data. The first example predicts dose response in cell lines based on drug and molecular features. The second example will show how to combine pathology from whole slide images and molecular features for cancer diagnosis and prognosis.

Acknowledgments: In addition to our speakers, we thank members of the Center of Biomedical Informatics and Information Technology (CBIIT) and BTEP who advised in organizing this series:

  • Dr. Amy Stonelake
  • Dr. Daoud Meerzaman
  • Dr. David Goldstein
  • Dr. Mariam Malik
  • Dr. Peter Fitzgerald

Questions? Contact the NCI Data Science Learning Exchange

Created by Clint Malone Last Modified Thu December 2, 2021 10:17 pm by Clint Malone