Cancerous tumors can be clinically described by their location in the body, size, and stage, and perhaps up to a dozen different attributes. But modern proteomics and genomics mean that nearly a million different attributes can now be extracted and used to characterize each tumor.

This generates enormous amounts of data. To begin to break down and make sense of this vast information store, a bioinformatics specialist, Bing Zhang, has been recruited to Baylor College of Medicine from Vanderbilt University School of Medicine with the help of a Rising Star Award from CPRIT. Having such large amounts of information about each tumor means doctors can begin to make better-informed decisions about patient care. But a main obstacle to using this information effectively is being able to integrate, manage, and interpret such vast volumes of data.

“Now we have tens of thousands of measurements on each tumor,” Zhang says, “and to derive better ways of knowing whether a tumor will recur or not we need to integrate all of these data.” Zhang says he also uses the data to create models to predict how certain tumors will respond to treatment, which can guide doctors’ decisions of how to treat each cancer based on integrated molecular data rather than solely on traditional clinical measurements.

Read More

Cancerous tumors can be clinically described by their location in the body, size, and stage, and perhaps up to a dozen different attributes. But modern proteomics and genomics mean that nearly a million different attributes can now be extracted and used to characterize each tumor.

This generates enormous amounts of data. To begin to break down and make sense of this vast information store, a bioinformatics specialist, Bing Zhang, has been recruited to Baylor College of Medicine from Vanderbilt University School of Medicine with the help of a Rising Star Award from CPRIT. Having such large amounts of information about each tumor means doctors can begin to make better-informed decisions about patient care. But a main obstacle to using this information effectively is being able to integrate, manage, and interpret such vast volumes of data.

“Now we have tens of thousands of measurements on each tumor,” Zhang says, “and to derive better ways of knowing whether a tumor will recur or not we need to integrate all of these data.” Zhang says he also uses the data to create models to predict how certain tumors will respond to treatment, which can guide doctors’ decisions of how to treat each cancer based on integrated molecular data rather than solely on traditional clinical measurements.

Proteomics and genomics information, integrated with clinical information about cancers can give doctors better information about whether the cancers need aggressive treatment or not. For example, some stage three colon cancers, even though they have already spread to the lymph nodes, will not recur because the tumors are not aggressive. On the other hand, about 20 percent of stage two colon cancers, which have invaded surrounding tissues but not spread to lymph nodes, will recur after surgery, Zhang says. “If a stage two is aggressive, we should treat it with chemotherapy, and we can use models to predict which tumor might benefit from such treatment.”

Zhang says his approach is based on a more systemic view of cancer genomics: it’s not single genetic mutations that produce a cancer phenotype. He believes that genomic information has to be integrated with information about what proteins are expressed in the tumors in order to have a complete view of the whole system. “Indeed, most targeted therapies are aimed at proteins in cancer cells,” he adds.

Zhang is developing a database of protein variants in cancer to eventually be able to use them as cancer biomarkers or therapeutic targets. He’s also making his integrated databases publicly accessible. “I was initially trained as a wet lab scientist,” Zhang says, “but I pursued additional training in bioinformatics during my postdoc, because I realized that data analysis is becoming the bottleneck in biological research.”

At Baylor, he appreciates his ability to collaborate directly with physicians involved in the clinical care of patients at the Texas Medical Center. He can also test his computational models directly in human tumors grown in mice. “We can predict that certain types of tumors will respond to a certain drug,” he says, “and we can now directly test that in mice.”

Zhang received his undergraduate education in biology at Nanjing University in China, and his Ph.D. in genetics from the Chinese Academy of Sciences in Shanghai. He was a postdoctoral fellow in bioinformatics and then a research scientist at Oak Ridge National Laboratory prior to joining the faculty at Vanderbilt in 2006.

Read Less