path-ai / hif2gene

Data and code to accompany our Nature publication

hif2gene: Dense, high-resolution mapping of cells and tissues from cancer pathology images for the interpretable prediction of molecular phenotypes

Overview

hif2gene is the primary code repository for reproducing analyses in Diao, Chui, and Wang et al. 2020: "Dense, high-resolution mapping of cells and tissues from pathology images for the interpretable prediction of molecular phenotypes in cancer"

You can read the full publication in Nature: Human-interpretable image features derived from densely mapped cancer pathology slides predict diverse molecular phenotypes

Installation

• Clone this repo to your local machine using https://github.com/Path-AI/hif2gene.git
• Estimated install time: <1 minute
• Estimated run time: varies between scripts (<1 minute to ~1 hour)

Contents

1. /scripts contains all Python and R code used to produce content in /figures from /data
2. /data contains all raw and cached data objects needed to reproduce the analysis from scratch
3. /figures contains all primary figures as individual vectorized .pdf

Version and package requirements

• R version: 3.6.2
  • caret
  • cluster
  • cowplot
  • data.table
  • devtools
  • dplyr
  • ggplot2
  • ggpubr
  • mclust
  • openxlsx
  • readxl
  • stringr
• Python version: 3.7.4
  • collections
  • copy
  • group_lasso
  • ipython
  • lifelines
  • math
  • matplotlib
  • numpy
  • pandas
  • pickle
  • plotly
  • random
  • scipy
  • seaborn
  • sklearn
  • statsmodels
  • sys
  • umap
  • warnings

Abstract

Computational methods have made substantial progress in improving the accuracy and throughput of pathology workflows for diagnostic, prognostic, and genomic prediction. Still, lack of interpretability remains a significant barrier to clinical integration. We present a novel approach for predicting clinically-relevant molecular phenotypes from whole-slide histopathology images using human-interpretable image features (HIFs). Our method leverages >1.6 million annotations from board-certified pathologists across >5,700 samples to train deep learning models for cell and tissue classification that can exhaustively map whole-slide images at two and four micron-resolution. Cell- and tissue-type model outputs are combined into 607 HIFs that quantify specific and biologically-relevant characteristics across five cancer types. We demonstrate that these HIFs correlate with well-known markers of the tumor microenvironment and can predict diverse molecular signatures (AUROC 0.601-0.864), including expression of four immune checkpoint proteins and homologous recombination deficiency, with performance comparable to ‘black-box’ methods. Our HIF-based approach provides a comprehensive, quantitative, and interpretable window into the composition and spatial architecture of the tumor microenvironment.

License

Shield:

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.