Data-Driven Discovery and Design of Synthetic Biological Circuits

This project aims to develop generalized design platforms, algorithms, and robust methods for reproducible experimental execution within the context of a distributed cloud lab. In short, we are interested in closing the loop between biological data and synthetic biological design.

The primary thrusts of this project involve: 1) the development of generalizable learning algorithms that discover governing dynamics of biological processes from experimental data generated by high-throughput automation or semi-automated workflows, 2) the development of explanatory network inference algorithms to explain failure modes or experimental surprise in synthetic biological circuits, and 3) design and planning algorithms to simultaneously evaluate, either in silico or in vivo, multiple experimental designs.

A major theme of this project is the fusion of data-driven models with biophysical insight. Data-driven learning algorithms can be used to discover models that faithfully recapitulate data, but they also need to yield or extract biophysical insight for improved biological design. This project centers questions on the appropriate design of experiments, post-identification analysis of models to extract meaningful insight, and modeling of circuits at scale.

This project is supported by the DARPA Information Innovation Office’s Synergistic Discovery and Design Program, lead by Dr. Jen Roberts, and is a collaborative multi-institutional program involving members from MIT, the Wyss Institute, Ginkgo Bioworks, Transcriptic, Netrias, Aptima, Two-Six Labs, Haverford College, the University of Washington, Duke University, Montana State University, and the Texas Advanced Computing Center at the University of Texas Austin.

Constructive Learning Algorithms for Microbial Pathogen Detection from Soil Samples

In this project, we aim to develop an end-to-end pathogen classification platform using a combination of advanced microbial extraction methods, multi-omics host-pathogen interrogation assays, and machine learning decision analytics. Currently, there is no standardized assay for quantifying or classifying the pathogenicity of a novel microbial organism derived from soil samples. This project aims to examine the morphological and genetic phenotypes that arise in host-pathogen interactions, to develop quantitative and categorical measures of pathogenicity.

Major research thrusts in this project include the development of optimized experimental assays to handle samples with extremely low cell counts, generating targeted mammalian reporter libraries for dynamic profiling of host-pathogen response, and the development of one-shot or few-shot learning algorithms with computational certificates of correctness.

This co-funded project is a collaboration with Drs. Janet Jansson, Becky Hess, Aaron Wright, and Rob Egbert at the Pacific Northwest National Laboratory, supported by the DARPA Biological Technologies Office “Friend or Foe” Program.

Machine-Assisted Design of Robust Biological Circuits with Extended Operational Envelopes

Synthetic biological circuits have, for the most part, been designed to operate in sustained rich media, in domesticated cell lines, at constant temperature and exponential growth phase. This project seeks to examine use a combination of high-throughput DNA assembly and machine learning to identify sequence determinants of synthetic biological parts that confer extended or complementary range of function in non-model growth conditions. How well do genetic circuits perform in the transition from exponential growth phase to stationary phase? Is it possible to engineer circuits that operate across the transition point? What are the fundamental limits of performance for genetic circuits optimized to work in stationary phase?

We seek to address these questions in the context of biologically-enabled surfaces and novel 3D printing materials. This work is supported by the Institute for Collaborative Biotechnologies, in collaboration with the Hawker, Begley, and Segalman Labs at UCSB, as well as US Engineer Research and Development Center (US-ERDC) and the Murray Biocircuits Group at the California Institute of Technology.

Genomic Optimization to Sustain Engineered Biological Function in Soil Microbes

This project seeks to develop methods and technological platforms for rapid high-throughput engineering of genomes of non-model microbes that exhibit robustness (and engineered fragility) to fluctuations in microbiome population dynamics and growth conditions.

The project is motivated by the observation that synthetic biological design is often delicate. To advance the field of synthetic biology to the next era of robust design, we seek to identify genetic and biological control mechanisms that confer increased or decreased fitness in organisms hosting synthetic gene circuits, either on plasmid or in the genome.

This project is a collaboration with Dr. Rob Egbert from the Earth and Biological Science Directorate with Pacific Northwest National Laboratory and Dr. Adam Deutschbauer with Lawrence Berkeley National Laboratory.

Discovering Fundamental Laws Governing Surface-to-Surface Transition in Prokaryotic Adaptation

This project seeks to blend experimental techniques of whole-cell measurement and high-dimensional sensor fusion of whole cell computational models that describe the adaptation response and learning mechanisms employed in surface-to-surface transition in hardy prokaryotic systems. What does the learning architecture of a bacteria suddenly transferred from soil to a river system look like? How do we build models and computational simulation engines that mimic the adaptation response?

This project is funded by an Army Research Office Young Investigator Program Award, running from 6/2020-6/2023

%d bloggers like this: