Listening to data isn’t easy. Massive amounts of data are often messy and complicated. But somewhere within the cacophony, information can harmonize and produce the sweet sound of discovery – if you have the right tools with which to hear it.
ATARiS is one of several tools developed at the Broad Institute to precisely tune in to the signals within datasets. The original idea for ATARiS came about a few years ago when members of Jill Mesirov’s computational biology and bioinformatics group, Bill Hahn's cancer biology group, and the Broad RNAi Platform were trying to address a common problem from the world of RNAi research. RNAi – short for RNA interference – allows researchers to “turn off” a gene or decrease that gene’s activity. Ideally, every gene in the genome would be paired with an RNAi reagent that could turn it – and only it – off. Instead, most RNAi reagents also disrupt other genes (a frustrating phenomenon known as off-target effects). Without a way to easily isolate on-target effects, the power of RNAi wanes.
RNAi is a critical tool for many projects at the Broad and beyond, including Project Achilles. This project – a joint effort between researchers at the Dana-Farber Cancer Institute and the Broad – seeks to pinpoint cancer’s most important weaknesses. To do so, researchers use RNAi to turn off genes in hundreds of cell lines. About 50,000 RNAi reagents have been used to target 11,000 of the 21,000 human genes (about five RNAi reagents for each of these genes) in order to see which genes are critical for cancer’s survival. These crucial genes could become the targets of drugs in the future.
“What we want to do is tune in on a specific target effect,” says Diane Shao, a graduate student in senior associate member Bill Hahn’s lab at the Broad Institute and Dana-Farber Cancer Institute. However, while researchers can pick out an RNAi reagent that seems particularly adept at killing cancer cells, they can’t be entirely certain which of its effects – on-target or off-target – are bringing about the desired result.
ATARiS helps cut through the noise from the multitude of variables and values. The computational method looks for patterns across multiple samples, assessing the performance of individual RNAi reagents to target specific genes. This allows researchers to determine which gene – rather than which RNAi reagent – is most of interest.
“ATARiS makes RNAi data more accessible,” says Aviad Tsherniak, a computational biologist in Jill Mesirov’s lab at the Broad and the key architect of ATARiS. “It simplifies it and standardizes it, and it makes the data compatible with other kinds methods.”