Date |
Nov 30, 2017 |
Speaker |
Jun Sese, Team leader of
AIST on machine learning (Visiting associate professor of
Bioinformatics Center)
|
Title |
Statistical significance of marker combinations: theory and applications
|
Abstract |
While the discovery of combinatorial markers is helpful for
personalized medicine, drug repositioning and more understanding of
molecular biology, the marker findings are not straightforward due to
limitations of computational and statistical methods. In this talk, I
first introduce that the problem comes from the difficulty to assess
statistical validity of the markers with the big-data analysis because
of multiple testing correction problem. Then, I propose a
statistically sound method to find the combinatorial factors named
LAMP, which counts the exact number of testable combinations and
calibrates the Bonferroni factor to the smallest possible value,
resulting in higher sensitivity than existing methods. The application
results of the method to GWAS data from Japanese large cohort
show the possibility to find novel maker combinations from existing data.
Also, the application to survival analysis suggests the novel relationships
between genes and markers.
Reference.
[1] Aika Terada, Mariko Okada-Hatakeyama, Koji Tsuda, and Jun Sese.
Statistical significance of combinatorial regulations.
Proc. Natl. Acad. Sci., Vol. 110, No. 32, pp. 12996-13001, 2013.
[2] Aika Terada, Ryo Yamada, Koji Tsuda, Jun Sese.
LAMPLINK: detection of statistically significant SNP combinations from
GWAS data.
Bioinformatics, 32 (22), 3513-3515, 2016.
[3] Raissa T Relator, Aika Terada and Jun Sese.
Identifying statistically significant combinatorial markers for
survival analysis.
GIW 2017
|
|