Machine learning (ML) is at the core of our strategy to accelerate drug discovery for novel cancer treatments, and scalable ML inference is at the core of our ability to rapidly analyze large molecular datasets. To scale up ML inference in a cost- and time-effective way, we built a parallel
Introduction At Reverie Labs, we focus on building state-of-the-art deep learning models to predict important molecular properties. One method we use in building such models is self-supervised pre-training, which we run over tens of millions of molecules. This requires overcoming two major hurdles: Working around peculiarities in how Tensorflow handles
At Reverie Labs, we use Python to power our machine learning model and data pipelines. We also use Python-based web frameworks such as Django for our internally deployed web applications (and for the less complicated ones, we use Streamlit!). In this blog post, we will be describing how we utilize
Reverie’s machine learning engineers are constantly building and improving our computational models to develop novel therapeutics. We often create Python tools to perform valuable computations such as filtering molecules for a particular characteristic or searching our internal database for specific PDB files to download. To make these scripts available
Exploration of Chemical Datasets using UMAP. Detailed analysis of how UMAP embeds a MoleculeNet dataset and practical uses for UMAP at Reverie Labs.
In this post, we’ll discuss how we manage and process PDBs, a critical chemical data filetype that encodes the 3-D structure of proteins. What is a PDB? At Reverie, we use 3D protein structural data for many applications - medicinal chemists evaluating structure-activity relationships, computational chemists running molecular dynamics
We're releasing a Colab notebook for training Transformer networks on a wide range of drug discovery tasks using Tensor2Tensor.
This blog post was originally published on the AWS Startups blog here. At Reverie Labs, we use computation to drive the development of therapeutics for cancer. To do this, we have built substantial cloud-based infrastructure to train machine learning models, deploy models to production, and build and ship internal-facing applications
Reverie scientists use our platform to run models with the ease of ordering online takeout.
A quick-and-dirty way to clean noisy datasets before training on them
In this blog, we’re going to be sharing some of the methods we use at Reverie to develop the models and high performance systems that power drug development programs. But first, we want to talk a little bit about why we are doing this. We started Reverie because millions