NEW YORK, July 24 - California-based startup ViaLogy, working with Intel, has created a new computational method that it said can greatly improve accuracy and sensitivity in microarray-data analysis.
Using an approach that relies on quantum-expressor functions to separate signal from noise, ViaLogy claims it can improve reproducibility in microarray experiments 100-fold, and reduce false positives from 30 percent down to as low as 1 percent.
The company's "active signal processing" method also allows researchers to detect minute microarray signals that are up to four orders of magnitude less than the amplitude of noise, says ViaLogy CEO and President Doug Lane.
"All the analyses done today evoke elaborate statistical measures to minimize the distance between signal and noise state," said Lane. "They try to filter out noise." By contrast, ViaLogy's approach models both noise and signal states and identifies perturbations in background noise.
Although the company has not yet published a research paper demonstrating the technique, Lane said that tests have shown ViaLogy's method can reduce inter-experiment variability in gene-expression studies from 1,500 to 2,000 genes down to only 10 to 15.
Pasadena-based ViaLogy relied on Intel's Itanium 2 processor to create its algorithms and is one of Intel's "family developer collaborators."
The company is currently talking with large, vertically integrated chip makers to whom it hopes to offer its software as part of a chip-analysis platform. ViaLogy also plans to market its software directly to researchers who make their own microarrays.
Lane says his company's analytic method solves one of the chronic problems of microarray experiments: the variability in results.
"[Microarrays] perform exceptionally well," he said. "The biochemistry in these devices is outstanding. The problems are strictly an artifact of the signal processing used."
ViaLogy is also developing algorithms for high-throughput sequencing, protein detection, and SNP analysis.