A gene analysis tool for cancer diagnostics
Dominique Lavenier and Patrick Durand
Developed by the GenScale research team at the Inria Rennes–Bretagne Atlantique centre, the Genome Assembly & Analysis Tool Box (GATB) is designed for data from next-generation gene sequencing. It currently includes a dozen software programs for various tasks in computational biology. The researchers plan to include a new application to facilitate diagnostics for cancer treatment in hospitals.
Over the past ten years, enormous progress has been made in genome sequencing. Its price is dropping before our eyes, so that DNA mapping may become as common as a blood sampling. This little revolution is explained by the arrival of higher-powered machines, known as next-generation sequencers (NGS). The systematic use of data from these devices will enable hospitals to improve how they target treatments, particularly for cancer. For this, however, the doctors responsible for the diagnostics must have new analytical software that can handle the enormous volumes of data and find the relevant information therein.
Search for mutations
To meet this need, the researchers of the GenScale bioinformatics research team are preparing to develop a new application. The idea? “To build a tool that will let the doctor observe in the patient that a certain number of genes which could mutate may be involved in a given type of cancer,” explains Dominique Lavenier , the team’s leader. “Using the observed mutations, the doctor can then focus the therapy and propose the chemical that is most appropriate for interacting with the targeted gene.” The new software will rely on the GATB library, a toolbox whose algorithms use little memory and work directly with compressed data. These characteristics constitute an advantage due to the great volume of genome data. “With the use of compact representation, our approach will save a great deal of time, both in terms of treatment and storage.” The oncologists at Rennes University Hospital are participating in the research. “They are the ones who guide us. We are currently in a detailed tool definition phase to ensure that it corresponds exactly to what is required. We would also like for it to be sufficiently generic that other hospitals in France will be interested.”
Making software widely available
To support this scientific research, Inria is currently funding a technology development initiative to make the software widely available. “We have a certain number of things that require further work with the doctors. Next, it will be up to me to transform the prototype into software that conforms to industry standards, ” notes Patrick Durand , research engineer responsible for writing the program. “In particular, the tool should comply with a certain number of regulations for software used for medical diagnostics. Afterwards, we’ll move onto a certification phase. ” For now, efforts are focused on defining the tool. One of the main questions concerns the establishment of criteria that enable the software to distinguish true mutations. “The sequencers tend to produce errors. The software must be capable of making the distinction between a true mutation and a simple sequencing error. This is in addition to natural polymorphisms, i.e., differences that can exist in an individual with regard to the reference human genome and which are also not mutations. ”
How to resolve these uncertainties? “Ideally, for each observed mutation, a level of probability must be assigned, ” responds Dominique Lavenier. “Then the doctor will then decide if there’s reason to investigate further. ” That said, the work of interpretation could also be facilitated. “If the software shows a mutation, it’s interesting to query the existing databases to know to what it corresponds and what the consequences are. This type of information already exists in the databases, but it’s scattered. We would like to bring it together so that, when the doctor observes a mutation that’s a little different, he can know where it has already been seen, the treatment that was used and the result. It is no longer only a tool that can show a mutation, but can also help interpret it. There is thus a large amount of information that can be used to add knowledge. It’s this type of decision-support tools that are currently lacking in hospitals. ”