New tools speed drug discovery and disease research

To study the genetic components of disease, researchers rely on mice or other research models in which particular genes are silenced, or turned off. In recent years, researchers discovered that they can selectively silence genes using small pieces of RNA called siRNA (short interfering RNA).

Unfortunately, sorting out which siRNA sequences block expression of which genes has proven to be truly daunting. Researchers at Whitehead Institute, however, recently released for public use a new computational tool that will vastly improve this process, streamlining drug discovery and disease research efforts.

A Needle in a Haystack

To harness the power of siRNA, researchers need predictive tools to narrow down the search for siRNA molecules that are most likely to affect a gene in a desired way without affecting the function of other genes. Until now, hunting for siRNA candidates has been like looking for a needle in haystack. Given a gene of 3000 nucleotides, there are 2980 possible siRNA candidates that might affect how the gene functions. (siRNA sequences are approximately 21 nucleotide bases long).

The Biocomputing Group at Whitehead has greatly simplified this process by devising and publishing a web-based tool that can quickly narrow down siRNA candidates.

“Scientists routinely came to Biocomputing asking how they could more efficiently predict siRNA targets,” says Lewitter. “Unfortunately, without suitable prediction tools, scientists had to randomly select siRNA strands from a pool of thousands of possibilities and hope that they would be successful in studying a gene of interest.”

Faced with scientists’ mounting frustration, Biocomputing took on the challenge to develop an easy and efficient tool to predict which siRNA molecules will be effective in a particular experiment. “Using this tool, researchers can narrow down the possibilities of potential siRNAs to a small handful that are likely to be effective for studying a particular gene,” says Lewitter.

To jumpstart the process, Lewitter initiated a collaboration with Tom Tuschl, a former Whitehead postdoc who had studied siRNA in David Bartel’s lab. Tuschl, who has further developed his characterization of siRNA, first at the Max Planck Institute and now at Rockefeller University, provided Biocomputing with a set of rules for determining candidate siRNAs devised from looking at many siRNA samples. These rules were based on qualities such as the size of the siRNA, its two-dimensional structure, and the components that start and stop the strands.

Using Tuschl’s rules, Bingbing Yuan of the Biocomputing Group wrote a series of computer programs that made these rules available to researchers through a simple web form. Users can enter the human or mouse genes that they are studying and specify certain criteria, and the program selects and displays potential siRNA sequences that can be used to generate a desired genetic effect. When the researcher selects a candidate, the program searches further. An email shortly appears in the user’s inbox with a weblink that shows a refined siRNA target sequence based on the candidate.

Taking it to the Bench

The web-based tool has proven to be a great resource for Whitehead scientists. For researcher David Sabatini, the web tool has streamlined his lab’s efforts to study genes that control cell growth. “Until recently, the process by which we identified siRNA candidates was arbitrary. We now have a tool that enables us to make more precise selections, which saves time, energy, and money,” says Sabatini.

Responding to the growing interest in siRNA from both academic and industry scientists, the Biocomputing Group has made their technology available to the public at http://jura.wi.mit.edu/bioc/siRNA/home.php. Based solely on word of mouth, groups in Europe and Asia, as well as throughout the United States, are already flocking to the site.

But, stresses Lewitter, this is just the tip of the iceberg. “Scientists face tremendous challenges in making use of today’s new technologies,” she says. “We’re trying to eliminate some of these hurdles by developing computational tools that make sense out of an otherwise overwhelming sea of data. Although we’ve made some great strides, it’s clear that we’ve only just begun.”

And so the work continues. Biocomputing’s current set of siRNA web tools will be followed by improvements based on examining all known siRNA experiments. The Biocomputing group is also collaborating with Carl Novina, a postdoc in Phillip Sharp’s lab in the Biology Department at MIT, who has developed an alternative method of predicting siRNA. Based on these contributions, Biocomputing intends to improve the predictive accuracy of the tool and provide a scoring system that will rank the efficacy of possible hits.