The Quest for CREST

Scientists working on the Pediatric Cancer Genome Project develop a powerful new tool to pinpoint the structural variations that give rise to childhood cancers.

by Elizabeth Jane Walker

The quest for CREST

Jinghui Zhang, PhD, began her journey to St. Jude Children’s Research Hospital as a preschooler perched high atop a stool in her mother’s immunology laboratory.

“My mom developed vaccines for children in China. As a kid, I would watch her doing experiments,” Zhang recalls. “I saw that by being a researcher you can save many lives.”

Today, Zhang uses her expertise in the computational aspects of genomics, systems biology and related computational sciences to help her colleagues identify the genetic changes that lead to childhood cancers. She and her team recently created a new tool that helps scientists accurately pinpoint cancer-causing mutations that occur within the 3 billion base pairs of DNA in the human genome.

The tool, dubbed Clipping Reveals Structure or CREST, was developed as part of the St. Jude Children’s Research Hospital – Washington University Pediatric Cancer Genome Project, which began early last year.

The search for variations

Scientists in the Pediatric Cancer Genome Project are sequencing normal and cancer cells from more than 600 childhood cancer patients. Investigators hope the three-year initiative will help them better understand the origins of childhood cancers and devise more effective ways to treat these diseases.

Soon after the project launched, scientists began sifting through data in search of structural variations—the chromosomal rearrangements and loss or amplification of DNA that lead to cancer. To find those structural variations, the team initially relied on several existing strategies, each of which had its shortcomings. Some methods yielded numerous false positives; others only indicated the general location of a structural variation, not the exact breakpoint, and missed key genetic lesions. At first, the researchers responded to these challenges by writing “filter” programs to remove the false positives.

“We thought those programs would be sufficient,” Zhang says.

Then, by accident, they discovered a gleaming needle in the genomic haystack. “When we were analyzing one of the leukemia samples, we stumbled upon an important cancer gene that existing tools had not detected,” Zhang recalls. “I realized that we needed to develop our own algorithm to make sure we don’t miss a lot of important hits in the cancer genome.”

The novel computational method the researchers developed identifies exactly when one copy of a normal gene moves to another chromosome. CREST has higher accuracy, precision and sensitivity than other strategies for finding structural variations.

“Other tools miss up to 60 to 70 percent of structural rearrangements in tumors,” Zhang says.

Soft clips, hard data

CREST uses pieces of DNA called soft clips to find structural variations.

Using a technique called next-generation sequencing, scientists break DNA molecules into small fragments, which are then copied and reassembled, using the normal genome as a comparison. Soft clips are segments of DNA that fail to align during that reassembly. Other scientists ignored the soft clips during the rearrangement process, but Zhang and her team used them to find insertions, deletions and inversions in the genomes.

The team tested their technique on a subtype of acute lymphoblastic leukemia (ALL) known as T-lineage ALL. In comparing the normal and cancer genomes of five St. Jude patients, the researchers discovered 89 new structural differences that they validated using other laboratory methods. To determine whether CREST also worked on adult cancers, the researchers applied the technique to a melanoma cell line that had already been sequenced and analyzed. In addition to finding 26 known structural variations, CREST uncovered 50 that had never before been identified.

“It has been really exciting to work on this,” Zhang says. “Through CREST, we were able to see that the rearrangements in the cancer genome can be very complex, involving many chromosomes and causing all sorts of disruptions that we never anticipated could happen. So this actually will provide a better understanding of how cancer initiates, evolves and progresses, and also how it leads to relapse.

“We’ve also found that the structural variation profiles appear to be quite different for different subtypes,” Zhang continues. “This will help with designing specific therapies that treat specific targets identifying these subtypes.”

Spreading the word

When the Pediatric Cancer Genome Project began in 2010, Dr. William E. Evans, St. Jude director and CEO, promised that the team would publicize its discoveries as they emerged.

“By doing that, we will be fulfilling the mission of St. Jude as well as our responsibility to lead,” he said. “And we will do so by sharing freely with the world what we discover.”

In keeping with that vow, Zhang and her colleagues recently published their results in the journal Nature Methods and placed the CREST instructions, user manual and test data online at

The little girl who began her career in an immunology lab thousands of miles away is now an integral part of a project that may eventually save the lives of children worldwide.

“I think the great thing about working at St. Jude is the close connection between the basic and clinical research,” Zhang says. “Working at St. Jude provides me with the opportunity to see that findings in basic research can turn into therapy. That’s extremely rewarding.”

Promise magazine, Autumn 2011

Donate Now Promise Archive

More articles from this issue