Publikace na Lékařská fakulta v Hradci Králové |
2014
Abstrakt
We describe a special regression algorithm for the identification of biomarker areas in SELDI-TOF mass spectra in this paper. „Tests in a set of orthogonal polynomial regressions“ is the basic principle of this approach. „Gnostic cluster analysis“ is then a very effective algorithmic complement especially for a case of excessive behavior of a part of (bio)markers. Apart from this another a new way of „TIC-normalization“ of data is proposed in this paper.
This new regression algorithm averages results significantly more effectively than software systems used. A very considerable amount of computation was made on a supercomputer.