with deep learning
play

Designing cancer vaccines with deep learning Mikhail Ignatov, Applied Mathematics and Statistics PhD candidate Why doesnt the immune system treat cancer? 1) Tumor microenvironment suppresses activity of immune cells 2) Tumor cells are

0 downloads 1 Views 430 KB Size Report
  1. Designing cancer vaccines with deep learning Mikhail Ignatov, Applied Mathematics and Statistics PhD candidate

  2. Why doesn’t the immune system treat cancer? 1) Tumor microenvironment suppresses activity of immune cells 2) Tumor cells are very similar to healthy cells We can force the immune system to target tumor cells using cancer immunotherapy

  3. How to target tumors? 1) Proteins in a cell are sliced into small pieces - peptides

  4. How to target tumors? 1) Proteins in a cell are sliced into small pieces - peptides 1) Peptides are loaded onto MHC molecules

  5. How to target tumors? 1) Proteins in a cell are sliced into small pieces - peptides 1) Peptides are loaded onto MHC molecules 1) MHC transport peptides to cell membrane and expose them to T-cells

  6. How to target tumors? 1) Proteins in a cell are sliced into small pieces - peptides 1) Peptides are loaded onto MHC Peptide molecules recognition 1) MHC transport peptides to cell membrane and expose them to T-cells 1) Upon recognition of the peptides T-cell can destroy the antigen presenting cell

  7. How to target tumors? 1) Proteins in a cell are sliced into small pieces - peptides 1) Peptides are loaded onto MHC If we know, which tumor derived peptides Peptide molecules bind to MHC, it is possible to develop recognition tumor-specific T-cells. But ... 1) MHC transport peptides to cell membrane and expose them to T-cells 1) Upon recognition of the peptides T-cell can destroy the antigen presenting cell

  8. The problem There are millions of different peptides in the cell. How to predict, which of them bind to MHC and become presented to the immune system?

  9. The problem There are millions of different peptides in the cell. How to predict, which of them bind to MHC and become presented to the immune system? Using Machine Learning

  10. Convolutional Neural Nets for binding prediction MHC Peptide

  11. Convolutional Neural Nets for binding prediction MHC Peptide Transform 3D model to an image

  12. Convolutional Neural Nets for binding prediction MHC Peptide Train the CNN to capture complicated interaction patterns Transform 3D model to an image

  13. Convolutional Neural Nets for binding prediction MHC Peptide Train the CNN to capture complicated interaction patterns Transform 3D model to an image Estimate binding probability

  14. Convolutional Neural Nets for binding prediction Why CNNs? 1) CNNs are highly relevant for image recognition tasks 1) A large number of frameworks exists for implementing CNNs, because of their high popularity 1) All the peptide-MHC interactions are highly similar in orientation, terminis are fixed 1) Sufficient data for training: ~400000 (non)interacting pairs 1) All current prediction tools rely solely on sequence data, without taking structural information into account. This results in very poor predictions for less frequent MHC variants and as a consequence in ineffective treatment

  15. How to create good quality models? Most of the training data exists only in sequence form We need to able to properly model peptide-MHC interaction ATFGHARFT

  16. Model ranking using SVM Modeling protocol: 1) Randomly sample multiple peptide conformations in MHC binding site 2) Minimize molecular mechanics energy function of each model 3) Sort models by similarity to the real conformation (RMSD) 4) Compute a large number of chemical features for each model 5) Train energy function specific for peptide-MHC interaction using SVM

  17. Pipeline 1) Model MHC-peptide interaction using SVM-based energy function 2) Predict binding score of the created model

  18. Timeline 1st year: Docking protocol design steps 1. Set up the testing framework for development of energy function (currently in development) 2. Train a number of SVM models using different feature sets (Rosetta energy terms) 3. Select the best SVM model for peptide-MHC docking CNN design steps 1. Prepare the data, create MHC-peptide models, choose training and testing sets 2. Train and test multiple CNNs with different architectures, pick the best one 3. Test the resulting CNN model on external datasets 2nd year: Study the possibility of adding patient’s T -cell repertoire. Learn to model T-cell interactions with peptide-MHC complex and introduce it into the prediction pipeline

  19. Thank you for attention

Recommend Documents


hao su
Hao Su July 6, 2017 Outline Overview

Deep 3D Representation Learning for Visual Computing Hao Su July 6, 2017

accelerate deep learning
ACCELERATE DEEP LEARNING WITH NVIDIA'S

ACCELERATE DEEP LEARNING WITH NVIDIA'S DEEP LEARNING PLATFORM | STEPHEN JONES |

deep learning on gpus
Deep Learning on GPUs March 2016 What

Deep Learning on GPUs March 2016 What is Deep Learning? GPUs and DL AGENDA

reproducibility and replicability in deep reinforcement
Reproducibility and Replicability in

Reproducibility and Replicability in Deep Reinforcement Learning (and Other

image classification with digits
Image Classification with DIGITS

Image Classification with DIGITS NVIDIA Deep Learning Institute 1 DEEP

deep neural networks and deep reinforcement learning
Deep Neural Networks and Deep

Deep Neural Networks and Deep Reinforcement Learning Deep Neural Networks

natural language processing with deep learning cs224n
Natural Language Processing with Deep

Natural Language Processing with Deep Learning CS224N The Future of Deep

an overview of deep residual learning
An Overview of Deep Residual Learning

An Overview of Deep Residual Learning Semih Yagcioglu 01.03.2016 Deep

deep reinforcement learning
Deep Reinforcement Learning [Mastering

Deep Reinforcement Learning [Mastering the Game of Go with Deep Reinforcement

mocha jl
Mocha.jl Deep Learning in Julia

Mocha.jl Deep Learning in Julia Chiyuan Zhang (@pluskid) CSAIL, MIT Deep

molecular determinants of factor viii immunogenicity
Molecular determinants of factor VIII

Molecular determinants of factor VIII immunogenicity Moderator Prof. dr.

cdc public health grand rounds
CDC PUBLIC HEALTH GRAND ROUNDS Changes

CDC PUBLIC HEALTH GRAND ROUNDS Changes in Clinical Diagnostics and Tracking

research day at the capitol
Research Day at the Capitol Poster

Research Day at the Capitol Poster Preparation Presentation Cody Bahavar

combining a soluble lag 3 protein with an anti pd 1
Combining a soluble LAG-3 protein with

Combining a soluble LAG-3 protein with an anti-PD-1 antibody in phase I-II

unlocking the potential of innovative medicines
Unlocking the potential of innovative

Unlocking the potential of innovative medicines Photochemical enhancement of

gastrointestinal devices for long t erm in situ delivery
Gastrointestinal Devices for Long-T

Gastrointestinal Devices for Long-T erm In Situ Delivery of Therapeutic

5529 adxs11 001 immunotherapy 12 month survival and
5529 ADXS11-001 IMMUNOTHERAPY: 12

5529 ADXS11-001 IMMUNOTHERAPY: 12 MONTH SURVIVAL AND SAFETY DATA FROM A

forget moonshots biomedicine needs an air traffic control
Forget Moonshots Biomedicine Needs an

Forget Moonshots Biomedicine Needs an Air Traffic Control System Jeff

the major histocompatibility complex
The Major Histocompatibility Complex

The Major Histocompatibility Complex Peter Burrows 2008 4-6529

particle cell interactions and their
PARTICLE-CELL INTERACTIONS AND THEIR

PARTICLE-CELL INTERACTIONS AND THEIR IMPLICATIONS IN LUNG DEFENSE MECHANISMS

cell communication
Cell Communication Topics 4.1 through

Cell Communication Topics 4.1 through 4.2 Topic 4.1 Cell Communication

immuno oncology applications
Immuno-Oncology Applications Lee S.

Immuno-Oncology Applications Lee S. Schwartzberg, MD, FACP West Clinic, P.C.;