Monica Berrondo pulls up a digital schematic of an antibody. It resembles bits of different-colored yarn stuck together by someone with no sense of design. Each color and bend represents information about the amino acids that comprise the antibody, explains Berrondo, whose company, Macromoltek, made this model possible. If the antibody shown in this image works, it may one day be deployed in the body of a sick person, where it will hunt down and cling to a specific pathogen. There, this colored squiggle may stimulate the immune system to go after the pathogen, mark it as an enemy for other immune system cells to take out, or attack it outright.
Antibodies, unlike many older pharmaceuticals, don’t have to carpet bomb healthy cells to attack specific pathogens. They’re efficient.
It used to take months to design an antibody like this, just to get it to the point where it could be created and tested. With Macromoltek it takes hours.
Researchers who subscribe to Macromoltek can log in to a secure workstation on the website and design a digital antibody using color-coded amino acids to read and track the sequence. The digital model helps researchers evaluate and edit their work, since one error can mess up the entire sequence. In general, Berrondo said, scientists will design maybe 10 antibodies that they hope will destroy a specific pathogen. They choose the best to go to a lab to be created and tested. But the digital design process significantly reduces both the time it takes to create an antibody, and the labor to test multiple iterations. Other companies have similar digital design capabilities, but Berrondo said Macromoltek is the only one she knows of that focuses on the creation of antibodies. The company has another distinction: it is the only one that lets researchers “humanize” antibodies that were originally created using mice.
User friendly science software
Berrondo is a graduate of Johns Hopkins University where she received her PhD in chemical and biomolecular engineering.
“I’ve always been a computer person,” she said. “I did a lot of internships in high school and I hated the applications. I didn’t find them very meaningful. So starting out in college I wondered ‘How can I find a software project that’s meaningful?’” She found the answer in computational biology and molecular modeling, using computers to help researchers solve medical problems.She also knew she wanted, one day, to start her own company. During her post-graduate work, while she was serving as a software consultant for laboratories, she got the inspiration for Macromoltek.
Most of the software laboratories were working with were difficult to use, for a lot of reasons. Some were too clunky and complicated, some too slow to do the calculations, some didn’t support object design, and so forth. Berrondo found herself returning to the same labs over and over to solve the same problems. So she launched into extensive market research to understand whether labs even used digital design—most didn’t—and if not why. She also asked the ones who did use it what issues they were having. Ultimately, she knew what she needed to make and spent six-to-nine months writing the software.
To find out if her software worked, Berrondo collaborated with the University of Texas on validation studies, designing antibodies and having them produced in the lab. With a dozen or so verified and another dozen in the works, the real world seems to be bearing out her software’s efficacy.
“Their protein modeling capabilities have provided us with a novel and reliable alternative to costly and occasionally challenging wet-lab experiments,” said molecular biologist Curt Hewitt of Signature Science, an Austin based subsidiary of Southwest Research Institute and Macromoltek customer. “Working with them has allowed us to focus on the bigger-picture scientific problems at hand with the confidence that the modeling and technical assumptions we’re operating under are solid.”
Last spring, Macromoltek became a Capital Factory Accelerator company. Since it was the accelerator’s first life science company, adviser Gordon Dougherty said he focused primarily on company strategy, customer acquisition, pricing and similar issues. Berrondo hired her mother, Susana Kaufman, a chemical engineer and software developer with experience in the financial services industry as her CFO.
“Monica is a quiet lion,” he said. “She’s reserved in her demeanor but unbelievably smart. When she asks a question, it’s a question with a purpose. I really respect that about her. I meet with a lot of entrepreneurs who shoot from the hip. She is different, in a good way.”
Though Berrondo has received grants, including recently receiving a National Science Foundation Phase II funding for $750,000 over the next two years, Dougherty said, she has come a long way without receiving investment money.
“She’s waiting until she proves things out to a certain level across her business plan. That’s hard to do. It takes a lot of patience.”
Now if she could design a program for that….
June 5, 2013
By SUSAN LAHEY
Reporter with Silicon Hills News
Kevin Nowka, Director of IBM Research, photo courtesy of Austin Forum on Science, Technology & Society
So here’s a sobering thought: According to the Sloan Digital Sky Survey there are 60 billion trillion stars in the observable universe. By the year 2017, according to Kevin Nowka, Director of Research for IBM, the total amount of data humans have generated will surpass that.
Much of that data, of course, is kitten memes.
Nowka spoke Tuesday at the Austin Forum on Science, Technology and Society held at the AT&T Executive Education and Conference Center on the topic: The Big Deal About Big Data.
According to the International Data Corporation, Nowka said, a fourth of the information in the digital universe would be useful for big data if it were tagged and analyzed but only three percent of useful data is tagged and even less than that analyzed.
Useful data includes data that could tackle the world’s problems such as wasteful medical spending, congested urban roadways, and environmental concerns. For example, the Institute of Medicine estimates of that of the $2.7 trillion spent on U.S. healthcare in 2011, $750 billion was wasted because of unnecessary and inefficiently delivered services, excessive administration costs, missed prevention opportunities and fraud. Some of which could be addressed with better data. Also, Nowka said, an IBM study revealed one in three business leaders frequently made decisions based on information they don’t trust or don’t have.
After sharing more than a dozen statistics about the growth of data streams—the volume of data doubles every year; and exploring the burgeoning number of sources—everything from social media to connected automobiles–Nowka introduced IBMs solution to problem of making use of it all: Watson Core Technology. This computer system uses natural language capabilities, hypothesis generation, and evidence-based learning to analyze and find answers in huge amounts of data. It is already being used by medical institutions such as Memorial Sloan Kettering who have trained the system to help diagnose and treat oncology patients and in experiments with retail and institutional banking organizations.
IBM is beginning to offer the technology on a limited basis to individual and business customers through the Watson Engagement Advisor Early Customer Program. The system still only speaks English and doesn’t deal well with images, Nowka said. But it can analyze natural language for subjects and verbs, dates, proper names, and learn as it goes, categorizing information and refining its hypotheses.
The Austin Forum is a monthly speaker series sponsored by the Texas Advanced Computing Center.