Harnessing information theory to find cancer genes

© iStock/Creative-Touch

A team of scientists have used a field of mathematics to help them uncover cancer genes.

The researchers at Johns Hopkins Medicine and Johns Hopkins Kimmel Cancer Center have applied information theory – a widely known field of mathematics designed to study how information is measured, stored, and shared – to look at genes that impact cancer. They say they have uncovered a likely key genetic culprit in the development of acute lymphoblastic leukaemia (ALL) – the most common form of childhood leukaemia.

The findings have been published in Nature Biomedical Engineering.

Applying information theory

The team used information theory, applying an analysis that relies on strings of zeros and ones — the binary system of symbols common to computer languages and codes — to identify variables or outcomes of a particular process. The scientists focused on a chemical process in cells called DNA methylation, discovered by a founder of the field of cancer epigenetics, Feinberg, in the 1980s, in which certain chemical groups attach to areas of genes that guide genes’ on/off switches. This is now recognised as a way of changing DNA with out altering its code.

Andrew Feinberg, M.D., M.P.H., Bloomberg Distinguished Professor at the Johns Hopkins University School of Medicine, Whiting School of Engineering and Bloomberg School of Public Health, said: “This study demonstrates how a mathematical language of cancer can help us understand how cells are supposed to behave and how alterations in that behaviour affect our health.”

Feinberg and his team say that using information theory to find cancer driver genes may be applicable to a wide variety of cancers and other diseases.

“Most people are familiar with genetic changes to DNA, namely mutations that change the DNA sequence. Those mutations are like the words that make up a sentence, and methylation is like punctuation in a sentence, providing pauses and stops as we read,” said Feinberg.

“We wanted to use this information to identify genes that drive the development of cancer even though their genetic code isn’t mutated,” said Michael Koldobskiy, M.D., Ph.D., paediatric oncologist and assistant professor of oncology at the Johns Hopkins Kimmel Cancer Center, who explained that methylation at a particular gene location is binary — methylation or no methylation — and a system of zeros and ones can represent these differences just as they are used to represent computer codes and instructions.

Analysing DNA

For the study, the team analysed DNA extracted from bone marrow samples of 31 children newly diagnosed with ALL at The Johns Hopkins Hospital and Texas Children’s Hospital, sequencing the DNA to determine which genes, across the entire genome, were methylated and which were not.

By assigning zeros and ones to pieces of genetic code that were methylated or unmethylated and using concepts of information theory and computer programs to recognise patterns of methylation, the scientists were able to find regions of the genome that were consistently methylated in patients with leukaemia and those without cancer.

They also saw genome regions in the leukaemia cells that were more randomly methylated, compared with the normal genome, a signal to scientists that those spots may be specifically linked to leukaemia cells. In particular, one gene, called UHRF1, stood out among other gene regions in leukaemia cells that had differences in DNA methylation compared with the normal genome.

“It was a big surprise to find this gene, as its link to prostate and other cancer has been suggested but never identified as a driver of leukaemia,” says Feinberg.

Experiments by the team show that laboratory-grown leukaemia cells lacking activity of the UHRF1 gene cannot self-renew and perpetuate additional leukaemia cells.

“leukaemia cells aim to survive, and the best way to ensure survival is to vary the epigenetics in many genome regions so that no matter what tries to kill the cancer, at least some will survive,” says Koldobskiy.

“This new approach can lead to more rational ways of targeting the alterations that drive this and likely many other forms of cancer.”

The Johns Hopkins team plans to use information theory to analyse methylation patterns in other cancers, and they will determine whether epigenetic alterations in URFH1 are linked to treatment resistance and disease progression in patients with childhood leukaemia.

Subscribe to our newsletter


Please enter your comment!
Please enter your name here