PORPOISE <img src="logo.png" width="75px" align="right" />

PORPOISE <img src="logo.png" width="75px" align="right" />

===========

===========

### Pan-Cancer Integrative Histology-Genomic Analysis via Multimodal Deep Learning

### Pan-Cancer Integrative Histology-Genomic Analysis via Multimodal Deep Learning

[Journal Link](https://www.cell.com/cancer-cell/fulltext/S1535-6108(22)00317-8) | [Interactive Demo](http://pancancer.mahmoodlab.org/) | [Graphical Abstract](https://www.youtube.com/watch?v=NnAaeGYUi_U)

[Journal Link](https://www.cell.com/cancer-cell/fulltext/S1535-6108(22)00317-8) | [Interactive Demo](http://pancancer.mahmoodlab.org/) | [Graphical Abstract](https://www.youtube.com/watch?v=NnAaeGYUi_U)

*TL;DR - We present an interpretable, weakly-supervised, multimodal deep learning algorithm that integrates whole slide images (WSIs) and molecular profile features for cancer prognosis. We validate our method on 14 cancer types, and extract both local and global patterns of morphological and molecular feature importances in each cancer type. Using the multimodal interpretability aspect of our model, we developed [PORPOISE](http://pancancer.mahmoodlab.org/), an interactive, freely-available platform that directly yields prognostic markers determined by our model for thousands of patients across multiple cancer types. To validate that these model explanations are prognostic, we analyzed high attention morphological regions in WSIs, which indicates that tumor-infiltrating lymphocyte presence corroborates with favorable cancer prognosis on 12 out of 14 cancer types.*

*TL;DR - We present an interpretable, weakly-supervised, multimodal deep learning algorithm that integrates whole slide images (WSIs) and molecular profile features for cancer prognosis. We validate our method on 14 cancer types, and extract both local and global patterns of morphological and molecular feature importances in each cancer type. Using the multimodal interpretability aspect of our model, we developed [PORPOISE](http://pancancer.mahmoodlab.org/), an interactive, freely-available platform that directly yields prognostic markers determined by our model for thousands of patients across multiple cancer types. To validate that these model explanations are prognostic, we analyzed high attention morphological regions in WSIs, which indicates that tumor-infiltrating lymphocyte presence corroborates with favorable cancer prognosis on 12 out of 14 cancer types.*

## Pre-requisites:

## Pre-requisites:

* Linux (Tested on Ubuntu 18.04) 

* Linux (Tested on Ubuntu 18.04) 

* NVIDIA GPU (Tested on Nvidia GeForce RTX 2080 Ti x 16) with CUDA 11.0 and cuDNN 7.5

* NVIDIA GPU (Tested on Nvidia GeForce RTX 2080 Ti x 16) with CUDA 11.0 and cuDNN 7.5

* Python (3.7.7), h5py (2.10.0), matplotlib (3.1.1), numpy (1.18.1), opencv-python (4.1.1), openslide-python (1.1.1), openslide (3.4.1), pandas (1.1.3), pillow (7.0.0), PyTorch (1.6.0), scikit-learn (0.22.1), scipy (1.4.1), tensorflow (1.13.1), tensorboardx (1.9), torchvision (0.7.0), captum (0.2.0), shap (0.35.0)

* Python (3.7.7), h5py (2.10.0), matplotlib (3.1.1), numpy (1.18.1), opencv-python (4.1.1), openslide-python (1.1.1), openslide (3.4.1), pandas (1.1.3), pillow (7.0.0), PyTorch (1.6.0), scikit-learn (0.22.1), scipy (1.4.1), tensorflow (1.13.1), tensorboardx (1.9), torchvision (0.7.0), captum (0.2.0), shap (0.35.0)

© [Mahmood Lab](http://www.mahmoodlab.org) - This code is made available under the GPLv3 License and is available for non-commercial academic purposes. 

© [Mahmood Lab](http://www.mahmoodlab.org) - This code is made available under the GPLv3 License and is available for non-commercial academic purposes.