Member Research and Reports

Member Research and Reports

Taiwan: Researchers Develop a Bayesian-based Algorithm to Identify Important Pathways/Genes Associated with Interested Outcomes

With the advancement in high-throughput genetic technologies, the cost and efforts have been greatly reduced to perform a genomewide screening of dysregulated pathways associated with interested outcomes. Several algorithms have been developed to select enriched pathways and pinpoint hub genes as potential regulators. However, challenge arises; those algorithms fail to take the crosstalk and compensation among pathways into consideration. These algorithms analyze each pathway as an independent subject and thus the biological meaning of cascading pathways disappears within the analysis. To address this issue, Dr. Tzu-Pin Lu and Dr. Chuhsing Kate Hsiao at National Taiwan University College of Public Health led the interdisciplinary research team to develop a Bayesian-based algorithm, which can concurrently analyze multiple pathways and report their relative importance based on posterior probabilities.

The study provided a systematic method to prioritize the importance of interested pathways. Notably, the simulation results showed that the sensitivity of the proposed algorithm outperformed other popular algorithms without inflating type 1 error rates. In a breast cancer study examined by the next-generation sequencing (NGS) technology, the proposed Bayesian algorithm has the highest chance to correctly identify associated and null pathways; while other popular algorithms may accidentally identify the null pathway significant.

In conclusion, through considering multiple biological pathways at the same time, the proposed Bayesian algorithm kept the advantage of selecting important pathways without losing the specificity to rule out insignificant one. The algorithm was developed by the R language and is available online.