第1232回生物科学セミナー

Assessing heterogeneity of data compared by correlation coefficients: theory and examples

Prof. Kay Diederichs(Department of Biology, University of Konstanz)

2018年09月28日(金)    14:00-15:00  理学部1号館東棟 380号室   

Data sets collected from different objects are often affected by systematic differences. In order to understand the changes, these differences need to be detected and analysed. An algorithm recently published [1] uses a multidimensional scaling approach of pairwise correlation coefficients of the individual data sets to separate random and systematic errors. As a result, systematic differences between datasets i,j are revealed by positions of the datasets in a unit sphere, at angles related to the amount of systematic deviation. Application to crystallographic datasets shows the presence of different clusters consisting of isomorphous datasets. Examples from other fields will be presented.

[1] Diederichs, K. (2017) Acta Cryst. D73: 286-293
https://doi.org/10.1107/S2059798317000699