PiCnIc produces maps of cancer progression in much the same way historical explorers drew maps of the Earth without satellite imaging. The program, like the explorers, draws tiny regional maps and then pieces them together like a puzzle.
The program extracts the simplest pieces of information from medical data, which Mishra calls "little blocks of causality." They describe one stage of cancer, along with a potential outcome.
Once the blocks of causality are constructed, PiCnIc examines them to retain only the likeliest paths. This process is akin to discarding poorly drawn regional maps before attempting to put together a global map.
The end result is a graphical representation of the possible scenarios a cancer patient might encounter. From it, physicians can hypothesize how cancer will progress and plan accordingly.
Just like a good cartographer, PiCnIc is capable of drawing all kinds of maps, as long as appropriate medical data are available. "[PiCnIc] is quite flexible. It can incorporate pretty much any cancer type," says Nicholas Navin, a professor at the University of Texas MD Anderson Cancer Center who wasn't involved in the creation of PiCnIc. Navin says that PiCnIc, at its current stage of development, "already gives [us] enough information to be clinically useful."
The program can also update the maps to factor in advances in cancer research as well, says Giulio Caravagna, a computational biologist at the University of Edinburgh who participated in the research project. As new medical data become available, researchers can rerun PiCnIc with ease to generate updated maps of cancer. "And there is already a pretty good amount of data, like the Cancer Genome Atlas," says Caravagna.
The research work was published on the Proceedings of the National Academy of Sciences in late June. Mishra says that he plans to look into "commercial opportunities" for PiCnIc and collaborations with "large research labs" to improve it.
For now, though, the PiCnIc maps "shouldn't be interpreted too much in making decisions for individual patient," says Navin.
"We are a little bit underpowered to comprehensively build these [maps of cancer]... there are still little ways to go," says Ben Raphael, a computational biologist at Brown University.