Mining statistically significant substrings using the chi-square statistic
Proceedings of the VLDB Endowment
| Hi-index | 0.00 |
Prion diseases (mad cow disease, CJD, etc.) are a groupof fatal neurodegenerative disorders associated withstructural conversion of a normal, mostly\alpha-helicalcellular prion protein (PrP) into a pathogenic \beta-sheet-richconformation. Little is known about which parts ofPrP undergo conformational transition and how diseaseassociated mutations facilitate this transition. In thiswork, we utilize a computational statistical approach todetect unusual patterns in prion protein. (i) We constructa novel entropic index which provides a quantitativemeasure of context-dependent conformational flexibilityof a sequence fragment. This index is used to studyconformational flexibility of PrP fragments. (ii) Weidentify PrP fragments that show unusual intrinsicstructural propensities. (iii) We estimate the statisticalsignificance of clusters of disease-associated PrPmutations using a stochastic model of mutational processwith unequal substitution rates and context-dependentmutational hot spots.