All News

Baltimore is among the epicenters of the nationwide substance abuse crisis. The city was recently named the “Overdose capital of the U.S” after a joint investigation by The New York Times and Baltimore Banner found it averaged 1,000 overdose deaths per year, a death rate nearly double of any other large American city.

In a concerted effort to curtail overdose deaths and improve treatment for people struggling with substance abuse, researchers at the University of Maryland (UMD) and University of Maryland, Baltimore (UMB) have partnered with a regional drug treatment center to apply machine-learning techniques to electronic medical record (EMR) data.

By using EMR to analyze data from patients including race, gender, location, traumas suffered, educational background and income, the cross-institutional team is hoping to identify trends that allow treatment to be better tailored to the individual and improve recovery rates.

“Our complex histories, individual abilities and family situations are all very personal to us, but there's going to be some commonalities among people suffering from addiction,” says Michael Cummings, a professor of biology with an appointment in the University of Maryland Institute for Advanced Computer Studies (UMIACS) who is one of the researchers on the project.

Cummings’ partners on the project include Edward Bernat, an associate professor of psychology at UMD, and Fadia Shaya, a Distinguished University Professor in the school of pharmacy at UMB. Yi Chen, a fifth-year doctoral student in UMD’s biological sciences program advised by Cummings, is also involved with the project.

Their work is supported by a $50,000 grant from the Institute for Clinical and Translational Research and the University of Maryland, Baltimore.

Bernat has extensively researched substance use disorders and its intersection with psychopathology. For the past few years, he has been working directly with Tuerk House— an inpatient drug center and crisis stabilization center in West Baltimore that sees 500–600 patients a month. Bernat provides ongoing clinical intake diagnostic services for the center’s clinical team.

Not long after he began conducting evaluations of the EMRs for patients arriving at Tuerk House, Bernat saw a need to involve machine learning to get the most out of the data he was viewing. This was due to the larger amount of data, from a larger number of patients.

“That’s where machine learning often has a big advantage—going through all of those different variables, and getting it crunched down into what’s actually meaningful with the minimal data available,” says Bernat.

The researchers plan to run machine learning algorithms on up to 10,000 of Tuerk House’s unique intake records from 2022 and 2023, with a focus on additional hurdles that Black substance users face that can impact their path to recovery.

African Americans complete substance abuse treatment only 40% of the time, which is 10% less than their White counterparts, according to a 2016 study published in the Journal of Substance Abuse Treatment. The study said that factors like socioeconomic status— specifically high school attainment and employment—are both associated with higher completion rates.

“If we find some subpopulations of people with certain characteristics are more at risk for relapse or something like that, then we can modify treatment for those sets of individuals.,” Cummings says.

Tuerk House’s partnership with UMD researchers for this project is part of a broader developing collaborative relationship to advance health disparities and treatment research relevant to Black substance users.

The inpatient drug center is in a majority black city that has been disproportionately damaged by the addiction epidemic, one that Chen is incredibly familiar with.

Chen moved to Baltimore to attend Johns Hopkins University, where he completed his B.A. in biology in 2014. For four years he did community service at Stepping Stone Ministry, a faith-based organization on the Hopkins campus that gave aid to the local homeless community, only stopping in 2020, when COVID–19 forced him to do so.

“Substance use disorder and homelessness are highly comorbid,” Chen says. “So, in some ways, this project is another way of serving the same population from that part of my life using a different set of skills.”

This project contrasts Chen’s previous work which has centered around molecular biological data sets, posing occasional challenges in refining and understanding data.

Chen noted the significance of the collaboration of experts from across various fields for this project.

“Substance use disorder is a complex disorder,” he says. “This project can hopefully address that by virtue of being a collaborative work involving psychopathology, social determinants, and machine learning.”

Looking ahead, Bernat envisions a dashboard wherein a patient completes the intake assessment at Tuerk House, and the system the research team is building will suggest individualized treatment plans based on the data received.

“Our local goal with is to develop something with [Tuerk House] clinicians that they can use quickly and efficiently,” he says.

Bernat and Cummings both emphasized that the overall goal is to understand different populations and improve the success rate of addiction treatment through a data-driven, personalized approach.

“That’s the ultimate benchmark we’re looking to achieve,” says Cummings. “Hopefully, data science and machine learning can provide some insights on how specialized health care can be improved and become more individualistic.”

— Story By Shaun Chornobroff, UMIACS communications group

The microbiome refers to the whole sum of microorganisms in a particular environment, such as the collective sum of gut bacteria in a human being. Microbiome research is a new frontier of scientific exploration. Studies that use big data technology to examine whole genomes of hundreds of organisms simultaneously represent a field called metagenomics. As this field matures, scientists are increasingly recognizing the need for sophisticated tools and technologies to decipher the complexities hidden within these microbial ecosystems.

To that end, on April 2, Mihai Pop, a professor in the Department of Computer Science and the director of the Institute for Advanced Computer Studies at the University of Maryland, gave a talk on the analytical challenges of microbiome science and how they can be combated by computational methods. The talk focused on the pivotal role of computational tools in unraveling the secrets of microbiomes and addressing the challenges associated with analyzing the vast datasets generated by these studies.

A key focus of metagenomics is the taxonomic classification of different microbes. The primary method for organizing and classifying microbes is comparing them to a database of known organism sequences. These similarity-based techniques are especially effective when the organisms in the sample are well represented in the database. Pop mentioned one of the most common similarity search methods used to classify microorganisms, the Basic Local Alignment Search Tool (BLAST). However, BLAST often misidentifies the closest neighbor to the microorganism of interest; the “most similar” organism according to BLAST may not actually be the most closely related.

“How can we find what’s the real [closest hit] if there is a hit? The E-value is misleading,” Pop explained during the talk, suggesting that BLAST may not always accurately identify the most similar organism to the microbiome of interest.

The E-values Pop mentioned refer to parameters in BLAST that describe the number of hits one can “expect” to see by chance when searching a database of a particular size. Pop also emphasized how many of these problems were only discovered years after BLAST integrated into common use.

“These are things we found out 20, 30, 40 years after [the computational tool] was written ... even though something has been used for many, many years, there [are] still things to learn about it,” Pop explained.

One of the other main challenges Pop highlighted is how the structure of biological databases affects scientists’ ability to reliably reveal insights on the microbiome. Reference databases are not all-encompassing. Many microorganisms cannot be cultured in labs, and a large proportion of those that can have not been sequenced or added to these reference databases. Consequently, not all environmental organisms are included in the sequence database, which limits the accuracy of similarity-based methods.

These problems are further compounded by the lack of contiguous information available in most sequencing datasets. Many sequencing analyses have to begin by joining together many sequence fragments and stitching together a whole related sequence. Assembling the sequencing data is also an unstandardized process, as new technologies used for assembling genomes are constantly being developed. These limitations can impede researchers' ability to derive meaningful insights and connections from microbiome datasets, because it substantially limits precision and decreases the accuracy of reference databases.

Pop then transitioned to discussing algorithms and software approaches to sequence similarity. Many current software used in classification employ the most recent common ancestor (MRCA) method. MRCA provides an annotation (marking of a specific feature of the DNA sequence) at the broadest taxonomic class that encompasses all of the possible markings in a sequence. However, this means that different types of software that use MRCA only make a few classifications at the genus or species level, meaning that stronger relationships between two microbes cannot be determined at the family, class or phylum level.

To address this challenge, Pop shared efforts from his own lab to develop advanced computational tools tailored specifically for microbiome analyses. He specifically focused on the Ambiguous Taxonomy eLucidation by Apportionment of Sequences (ATLAS). ATLAS is a data-driven database partitioning method, which aims to divide a large dataset into smaller, more easily analyzable datasets. ATLAS groups sequences into biologically meaningful partitions by querying the sequence against a reference database and then identifying and clustering hits that are considered significant. ATLAS also represents a shift away from the MRCA method.

As the talk concluded, Pop emphasized the critical need for interdisciplinary collaboration to advance microbiome research. Integrating expertise from fields such as biology, computer science and statistics is essential for developing innovative solutions to microbiome-related challenges. This interdisciplinary approach allows researchers to harness the power of computational tools to extract meaningful patterns and associations from microbiome datasets.

—This article by Shreya Tiwari was originally published in The News-Letter, a student publication at John Hopkins University.

Jisha Jesudass (right in photo) is responsible for designing and maintaining a sophisticated network system used by UMIACS faculty to seamlessly move data among various stakeholders both on and off campus.

For almost four decades, a powerful data center in the A.V. Williams Building has been an essential part of the University of Maryland Institute for Advanced Computer Studies (UMIACS), supporting the institute’s mission of incentivizing cutting-edge research and staying at the forefront of technological innovation.

As the institute’s research portfolio continues to expand, the infrastructure needed to support it—particularly in compute-heavy areas like artificial intelligence (AI) and machine learning—also needs to grow.

In April, UMIACS data center staff began coordinating a three-month renovation to expand the center’s physical footprint by 50 percent, growing from 2,300 square feet to 3,450 square feet.

The expansion project will provide additional space for computing hardware and will offer better options to ensure that the data center’s power grid and cooling systems can handle any increased workload, says Derek Yarnell, the institute’s director of computing facilities.

Much of the current workload is handled by multiple racks of graphical processing units (GPUs), powerful parallel computing hardware used for gaming, content creation—and in the case of UMIACS—machine learning and neural networks for high performance computing in areas like AI and computer vision.

In the past five years, UMIACS has increased the number of GPUs in its data center by almost tenfold, Yarnell says. This has allowed researchers to perform certain tasks at almost one petaFLOP of speed—equivalent to more than one quadrillion floating-point operations per second.

“If you measure the amount of computational infrastructure in place that is dedicated specifically toward AI and machine learning research, we are the largest on campus by a wide margin,” Yarnell says. “And we believe this is only the beginning.”

UMIACS faculty are constantly seeking new mechanisms to upgrade their computing resources, says Tom Goldstein, a professor of computer science and director of the University of Maryland Center for Machine Learning (CML).

Goldstein advocates for making economy of scale purchases, wherein CML faculty collectively pool portions of their individual research grants to purchase racks, routers, wiring and accessories needed to support multiple GPU clusters.

The data center expansion matches well with that concept.

“We want to be forward-looking in regard to increasing our computing capabilities, particularly those of us requiring high-memory nodes for projects that focus on large language models and other emerging AI research,” Goldstein says.

Other UMIACS faculty also want to improve both the availability and level of computing power available through the institute’s data center.

Furong Huang, an assistant professor of computer science with an appointment in UMIACS, is working with her graduate students to build a foundation model for sequential decision making, a form of generalist AI that holds the capacity to learn and master diverse sets of tasks downstream.

“Imagine an AI that not only learns quickly, but adapts to your daily tasks autonomously,” says Huang, who is a member of CML.

This work requires analyzing massive amounts of data used in the AI system, a task best handled by a large network of multiple GPUs, Huang says.

Supporting these myriad research efforts falls on Yarnell’s team, a dedicated staff of network engineers and high-performance computing experts assisted by more than 20 undergraduates working at the UMIACS Help Desk.

The institute’s technical staff are involved in the computing process from start to finish, Yarnell says. This includes helping faculty spec out their equipment needs to submit in their research proposals; working with UMD procurement to purchase the equipment; installing all the hardware and software when it arrives on campus; and maintaining the equipment over its technical lifespan by providing security patches and updates.

The tech staff is also responsible for managing and securely storing all the data used in UMIACS research. Current storage capabilities top out at around 3.65 petabytes, with that number constantly increasing, says UMIACS network engineer Jisha Jesudass.

There has also been steady growth in the network capabilities for the institute. When UMIACS moved from the A.V. Williams Building to the Brendan Iribe Center for Computer Science and Engineering in 2019, Jesudass’ job responsibilities included designing and maintaining a high-capacity network—miles of wiring and dozens of sophisticated routers—used to transmit large volumes of data both on and off campus.

She emphasized that the upgrades currently underway in the data center include further improving connectivity capabilities to make the workflow of UMIACS projects more efficient.

“Every job or every researcher right now, they want to minimize latency and delays. This new infrastructure will help a lot with that,” Jesudass says.

For Yarnell (pictured left), the renovation and expansion of the data center remains a work in progress and a labor of love. His involvement with the data center started in 1998 as an undergraduate working at the UMIACS Help Desk. Since that time, he’s seen—and helped lead—much of the data center’s growth.

“Our data center has obviously stood the test of time up to this point—it’s like a Ferrari, a fine-tuned machine that has aged very well,” he says. “But we still think we can continue to innovate it, and continue to build it, in a way that will allow the center to support any new opportunities that are yet to come.”

—Reporting on this story by Shaun Chornobroff, UMIACS communications group

Researchers at the University of Maryland and National Institutes of Health have identified the microbial enzyme responsible for giving urine its yellow hue, according to a new study published in the journal Nature Microbiology on January 3, 2024.

The discovery of this enzyme, called bilirubin reductase, paves the way for further research into the gut microbiome’s role in ailments like jaundice and inflammatory bowel disease.

“This enzyme discovery finally unravels the mystery behind urine’s yellow color,” said the study’s lead author Brantley Hall, an assistant professor in the University of Maryland’s Department of Cell Biology and Molecular Genetics. “It’s remarkable that an everyday biological phenomenon went unexplained for so long, and our team is excited to be able to explain it.”

When red blood cells degrade after their six-month lifespan, a bright orange pigment called bilirubin is produced as a byproduct. Bilirubin is typically secreted into the gut, where it is destined for excretion but can also be partially reabsorbed. Excess reabsorption can lead to a buildup of bilirubin in the blood and can cause jaundice—a condition that leads to the yellowing of the skin and eyes. Once in the gut, the resident flora can convert bilirubin into other molecules.

“Gut microbes encode the enzyme bilirubin reductase that converts bilirubin into a colorless byproduct called urobilinogen,” explained Hall, who has a joint appointment in the University of Maryland Institute for Advanced Computer Studies and is a core faculty member in the Center for Bioinformatics and Computational Biology. “Urobilinogen then spontaneously degrades into a molecule called urobilin, which is responsible for the yellow color we are all familiar with.”

Urobilin has long been linked to urine’s yellow hue, but the research team’s discovery of the enzyme responsible answers a question that has eluded scientists for over a century.

Aside from solving a scientific mystery, these findings could have important health implications. The research team found that bilirubin reductase is present in almost all healthy adults but is often missing from newborns and individuals with inflammatory bowel disease. They hypothesize that the absence of bilirubin reductase may contribute to infant jaundice and the formation of pigmented gallstones.

“Now that we’ve identified this enzyme, we can start investigating how the bacteria in our gut impact circulating bilirubin levels and related health conditions like jaundice,” said study co-author and NIH Investigator Xiaofang Jiang. “This discovery lays the foundation for understanding the gut-liver axis.”

In addition to jaundice and inflammatory bowel disease, the gut microbiome has been linked to various diseases and conditions, from allergies to arthritis to psoriasis. This latest discovery brings researchers closer to achieving a holistic understanding of the gut microbiome’s role in human health.

“The multidisciplinary approach we were able to implement—thanks to the collaboration between our labs—was key to solving the physiological puzzle of why our urine appears yellow,” Hall said. “It’s the culmination of many years of work by our team and highlights yet another reason why our gut microbiome is so vital to human health.”

###

This article was adapted from text provided by Brantley Hall and Sophia Levy.

In addition to Hall, UMD-affiliated co-authors included Stephenie Abeysinghe (B.S. ’23, public health science); Domenick Braccia (Ph.D. ’22, biological sciences); biological sciences major Maggie Grant; biochemistry Ph.D. student Conor Jenkins; biological sciences Ph.D. students Gabriela Arp (B.S. ’19, public health science; B.A. ’19, Spanish language), Madison Jermain, Sophia Levy (B.S. ’19, chemical engineering; B.S. ’19, biological sciences) and Chih Hao Wu (B.S. ’21, biological sciences); Glory Minabou Ndjite (B.S. ’22, public health science); and Ashley Weiss (B.S. ’22, biological sciences).

Their paper, “Discovery of a gut microbial enzyme that reduces bilirubin to urobilinogen,” was published in the journal Nature Microbiology on January 3, 2024.

This research was supported by the NIH’s Intramural Research Program, the National Library of Medicine and startup funding from UMD. This article does not necessarily reflect the views of these organizations.