Stanford Medicine Newsletter Updates For the Local Community

 

Michael Levitt, PhD

Merging biology and computation

Levitt accepts his prize at the Dec. 10 ceremony at the Stockholm Concert Hall.

   

Early on, you helped bring together the worlds of computation and biology. Why is this marriage important, and how has it advanced the learning curve in biomedicine?

Computers have become essential aids to help understand, predict, modify and design complicated systems. Proper understanding and manipulation of biomedically important proteins and large nucleic acid molecules responsible for the intricate working of life cannot rely simply on experimental work. Increasingly scientists must use computer simulations like those we pioneered from 1967 to 1976.

Key to our work was the design of multiscale models for biological macromolecules. These models were simple enough to enable calculations on machines 10 million times less powerful than today’s computers yet not so simple that they missed the essential details. Somewhat surprisingly, the same models have endured over the past 46 years and are still used in the most advanced simulations today.

Your work has enabled scientists to make computer models of molecules and molecular processes. How might that help in understanding disease?

Diseases and their treatment all depend on the physical and chemical properties of molecules. Any method to improve models of these molecules has huge implications for biomedical science in general and for human health in particular. Drugs will increasingly be designed by computer methods.

One example in my own work involves the use of antibodies to treat cancer. An obvious way to treat cancer would be to inject cancer cells into a healthy patient, harvest the antibodies made against these disease cells and then inject the purified antibodies into the patient. This is clearly out of the question for ethical reasons. Instead, anti-cancer antibodies are raised in mice. The antibody recognizes the cancer cells but is seen as foreign by human cells and is rejected. The computer must model the mouse antibody and then change it to be more like a human antibody (a process called humanization).

Shortly after joining Stanford in 1987, I consulted for a local startup and modeled antibodies for them on the computer. This led to a key paper in 1989 and a patent issued that same year.

The method was effective in designing an antibody that was both effective against the cancer cells and well tolerated by the patient. It led to an industry with annual sales of many billions of dollars and annual company royalties of $400 million. The patent expires in one year, but this story indicates how calculations, together with protection of intellectual property, years of manufacturing efforts and marketing skills, can lead to massive advances.

We have entered the era of so-called big data, in which scientists can use vast repositories of biomedical data to develop innovative approaches to treatment. What will be the impact of big data?

The phenomenon of big data is both an opportunity and a threat. In the past, data accumulated slowly, which allowed the data to be examined manually, and the brain was used to find underlying patterns. It is key that the patterns be based on some underlying understanding of the system, such as physical laws. Today big data is often approached by statistical methods and computer machine-learning techniques that look for correlations without understanding the physical basis for such effects. This is easily done but may lead to misleading conclusions.

Can you compare the computers you used in the early days of your research in the late 1960s with those available today?

The Golem computer built at the Weizmann Institute in Israel for about $5 million in U.S. dollars ($35 million today) did a calculation on the protein lysozyme in 18 minutes in 1968. Today the same calculation on my laptop costing $3,500 takes 1/10 of a second. This means that the calculation is 10,000 times faster on a computer costing 10,000 times less. The Golem computer filled a large hall, whereas my laptop fits into a large shoebox.

Such huge changes are hard to comprehend, so imagine if motorcars had changed the same way. A Cadillac that cost $6,000 in 1965 ($40,000 today) would now cost just $4. More amazingly, it would have a top speed of a million miles an hour and be able to carry 50,000 people.

Is there one vivid memory that stands out from the Nobel ceremony in Stockholm?

My most vivid memory is the cold sweat that overcame me minutes before I had to give my after-dinner speech. I was terrified, as I had decided to start out with 34 words in Swedish, a language I cannot read, write or speak. “Ers Majestäter, Ers Kungliga Högheter, Ers Excellenser … ,” I said: “Your Majesties, Your Royal Highnesses, Your Excellencies, ladies and gentlemen, I start in Swedish to prove that I can still learn something 45 years after doing the work that brought me here.”

.

Stanford Medicine Resources:

Footer Links: