Awesome

Peax: a pattern explorer for epigenomic data

Peax's UI

Epigenomic data expresses a rich body of diverse patterns, but extracting patterns genome wide is limited, as specialized algorithms are required or the expressiveness is low. Peax is a tool for interactive concept learning and exploration of epigenomic patterns based on unsupervised featurization with autoencoders. Genomic regions are manually labeled for actively learning feature weights to build custom classifiers based on your notion of interestingness.

Installation

git clone https://github.com/Novartis/peax peax && cd peax
make install

Do not fear, make install is just a convenience function for setting up conda and installing npm packages.

Note: If you're a macOS user you might need to brew install libpng and openssl for the pybbi package. Also see here.

Overview

Peax consists of four main parts:

A python module for creating a autoencoder. [/ae]
A set of example notebooks exemplifying the creating of autoencoders. [/notebooks]
A Flask-based server application for serving genomic and autoencoded data on the web. [/server].
A React-based user interface for exploring, visualizing, and interactively labeling genomic regions. [/ui].

Getting started

Quick start

Peax comes with 2 example autoencoders for which we provide convenience scripts to get you started as quickly as possible.

ChIP-seq encoder for 12Kb genomic windows at 100bp binning.
```
make example-12kb
```
DNase-seq encoder for 2.4Kb genomic windows at 100bp binning
```
make example-2_4kb
```

The convenience scripts will download test ENCODE tracks and use the matching configuration to start the server.

Note: if you need to run Peax on a hostname or port other than localhost:5000 you need to manually start Peax using the respective configs which you can find in the Makefile.

Slow start

Build an autoencoder

Note: right now only Keras-based autoencoders are supported.

First you need to create an autoencoder for your datasets's content types. The design of the autoencoder is entirely up to you. To get started take a look at the notebooks that we provide.

After creating your autoencoder save its model as an HDF5 file.

Configure Peax with your data

Next you need to configure Peax with your data. The main reason for doing is to tell Peax, which tracks you want to visualize in HiGlass and which of those tracks are encodable using an autoencoder.

The fastest way to get started is to copy the example config:

cp config.json.sample config.json

The main part to adjust is aes and datasets. AE stands for autoencoder and is a list of autoencoder definitions, e.g., you could have multiple encoder for different datatypes. The required format for autoencoders is as follows:

Field	Description	Defaults	Dtype
encoder	Relative path to your pickled encoder model. (hdf5 file)		str
decoder	Relative path to your pickled decoder model. (hdf5 file)		str
content_type	Unique string describing the content this autoencoder can handle. Data tracks with the same content type will be encoded by this autoencoder.		str
window_size	Window size in base pairs used for training the autoencoder.		int
resolution	Resolution or bin size of the window in base pairs.		int
latent_dim	Number of latent dimensions of the encoded windows.		int
input_dim	Number of input dimensions for Keras. For 1D data these are 3: samples, data length (which is `window_size` / `resolution`), channels.	3	int
channels	Number of channels of the input data. This is normally 1.	1	int

Example:

{
  "encoder": "path/to/my-12kb-chip-seq-encoder.h5",
  "decoder": "path/to/my-12kb-chip-seq-decoder.h5",
  "content_type": "chip-seq-pval",
  "window_size": 12000,
  "resolution": 100,
  "channels": 1,
  "input_dim": 3,
  "latent_dim": 12
}

Datasets require the following format:

Field	Description	Dtype
filepath	Relative path to your track data file (bigwig or bigbed).	str
content_type	Unique string describing the content this dataset. If you want to search for patterns in this track you need to have an autoencoder with a matching content type.	str
uuid	A unique string identifying your track. (Optional)	str
name	A human readable name to be shown in HiGlass.(Optional)	str

Example:

{
  "filepath": "data/chip-seq/my-fancy-gm12878-chip-seq-h3k27c-track.bigWig",
  "content_type": "chip-seq-pval",
  "uuid": "my-fancy-gm12878-chip-seq-h3k27c-track",
  "name": "My Fancy GM12878 ChIP-Seq H3k27c Track"
}

Start Peax

Finally, start the Peax server to run the application:

./start.py --config=your-config.json

Start supports the following options:

usage: start.py [-h] [-e ENCODER] [-d DATASET] [-w WINDOWSIZE] [-r RESOLUTION]
                [-s STEPSIZE] [-c CHROMS] [--config CONFIG] [--clear]
                [--debug] [--host HOST] [--port PORT] [--verbose]

Peak Explorer CLI

optional arguments:
  -h, --help            show this help message and exit
  -e ENCODER, --encoder ENCODER
                        path to saved encoder
  -d DATASET, --dataset DATASET
                        path to saved dataset (bigwig)
  -w WINDOWSIZE, --windowsize WINDOWSIZE
                        path to saved dataset (bigwig)
  -r RESOLUTION, --resolution RESOLUTION
                        number of bp per bin
  -s STEPSIZE, --stepsize STEPSIZE
                        relative to window, e.g., `2` => `windowsize / 2 =
                        stepsize in bp`
  -c CHROMS, --chroms CHROMS
                        comma-separated list of chromosomes to search over
  --config CONFIG       use config file instead of args
  --clear               clears the db on startup
  --debug               debug flag
  --host HOST           Customize the hostname
  --port PORT           Customize the port
  --verbose             verbose flag

The hostname defaults to localhost and the port of the backend server defaults to 5000.

Development

Handy commands to keep in mind:

make install installs the conda environment and npm packages and builds HiGlass
make update updates the conda environment and npm packages and rebuilds HiGlass
make build rebuild the peax ui
./start.py starts the Flask server application for serving data
[/ui]: npm install installs and updates all the needed packages for the frontend
[/ui]: npm build creates the production built of the frontend
[/ui]: npm start starts a dev server with hot reloading for the frontend

To start developing on the server and the ui in parallel, first start the backend server application using ./start.py and then start the frontend server application from ./ui using npm start. Both server's watch the source code, so whenever you change something the servers will reload.

Configuration

There are 2 types of configuration files. The backend server configuration defines which datasets to explore and is described in detail above.

Additionally, the frontend application can be configured to talk to a different backend server and port if needed. Get started by copying the example configuration:

cd ui && cp config.json.sample config.json

By default the server is dynamically set to the hostname of the server running the frontend application. I.e., it is assumed that the backend server application is running on the same host as the frontend application. The port of the server defaults to 5000.

Start the backend and frontend apps

For development the backend and frontend applications run as seperate server applications.

# Backend server
./start.py --config=config.json --debug

# Frontend server
cd ui && npm start