Skip to main content
Premium Trial:

Request an Annual Quote

Weill Cornell Aims For 'Precision Metagenomics' With Quantitative Computing


CHICAGO (GenomeWeb) – Get ready for a future of "precision metagenomics" and "geospatially informed medical care."

Those are two of the goals of the newly formed WorldQuant Initiative for Quantitative Prediction at Weill Cornell Medicine in New York City. Weill Cornell announced the program's formation two weeks ago, courtesy of a $5 million gift from quantitative investment firm WorldQuant and its founder Igor Tulchinsky.

The initiative seeks to develop and apply predictive tools and quantitative methods to help researchers and clinicians better understand the genetic factors that drive disease in individual patients.

Physician-scientists at the Caryl and Israel Englander Institute for Precision Medicine and the Sandra and Edward Meyer Cancer Center at Weill Cornell Medicine will collaborate with WorldQuant professionals to analyze clinical samples and visualize various diseased tissues at single-cell resolution. They will lean on WorldQuant's supercomputing infrastructure, which will include new software that applies advanced pattern-recognition algorithms to model disease progression, the organizations said.

"We are looking to develop diagnostics methods that combine single-cell omics and single-molecular measurements from tissue and blood with longitudinal patient health records, and then use machine learning methods such as deep learning to predict how diseases such as cancer may develop, respond to therapy, and evolve in time, [and] whether and when side effects, toxicities, or resistance may occur upon treatment," Christopher Mason, co-director of the WorldQuant Initiative for Quantitative Prediction at Weill Cornell Medicine, said in an email to GenomeWeb.

"When available, we will leverage other types of longitudinal data such as those from wearables and cell phones, as well as home and personal data. By applying these methods to large numbers of clinical samples with longitudinal patient health records and focusing on metrics quantifying tumor heterogeneity and diversity of the tumor microenvironment, we will learn how to connect these metrics with outcome and most importantly to build predictive models of clinical outcomes," Mason continued.

They will perform single-cell proteomics imaging with mass cytometry, single-cell RNA sequencing, single-cell epigenomics, genetic and epigenetic analysis of cell-free nucleic acids, and develop new metagenomics methods, according to Mason and his co-director of the initiative, Olivier Elemento. They also will explore ways to translate WorldQuant algorithms and data models — originally built to predict fluctuations in the stock market — to cancer research.

Mason called the World Quant Initiative for Quantitative Prediction "a unique effort to combine technologies that make measurements at single-cell and single-molecule resolution with advanced machine learning to build predictive models of disease risk and disease evolution."

On the clinical side, the initiative will concentrate on optimizing measurements for clinical samples and building predictive models of disease evolution and response to treatment, particularly in cancer and metagenomics, Elemento said.

The Weill Cornell team initially will apply these technologies to "tumor types with poor outcomes," Mason said, naming leukemias, lung tumors, certain types of breast cancer, as well as microbiome/metagenome samples.

"The predictive models will be learned from clinical samples and applicable prospectively to clinical samples where the outcome is not known. Once validated, these assays may be directly applicable in the clinic," said Elemento, associate director of the Institute for Computational Biomedicine at Weill Cornell. Later, the researchers may consider integrating such models into the electronic health record to support real-time health analytics.

"Another compelling aspect of the initiative is the creation of 'precision metagenomics,' where we use cross-kingdom algorithms to discern the impact of the microbiome and metagenome on cancer risk, as well as disease or infection progression, leveraging our creation of large-scale, global metagenomics datasets like MetaSUB," Mason said. "This will help us create 'geospatially informed medical care,' since we know the GPS coordinates, type, and city-specific density of the world's antibiotic-resistance genes."

Weill Cornell Medicine plans on eventually releasing to other medical institutions some of the algorithms it develops with WorldQuant. "We think that every person can benefit from more quantitative and predictive medicine, from each unique genome of NASA's astronauts before flight to every pensive patient that walks into our hospital doors," said Mason.

Tulchinsky said WorldQuant made the donation because Weill Cornell and his company have similar approaches to applying technology to analyze data. "We are both excited about the potential novel application of algorithms and models that can be used in the medical field," Tulchinsky said in a separate email.

"We are in the age of prediction, and I believe that data and technology will be the driving force in solving complex problems, today and in the future. I could not be more pleased to be a partner in this program by leveraging WorldQuant’s computing and quantitative capabilities to drive advances in predicting the future risk of developing disease and treatment of various illnesses, including cancer," he added.

The two organizations are already very familiar with each other. Tulchinsky is a member of the board of overseers at Weill Cornell Medicine, while Mason, a geneticist and computational biologist, has won a WorldQuant Foundation Research Scholar Award.