Algorithmsbasedonprobability theorycanaddressissues of uncertainty directly throughtheir representational frameworkandtheir theoryfor data combination.In this paper, wediscuss the advantages of probabilistie formulationsfor molecular-structurecalculations,describe oneimplementationof such a formulation, and showits performanceon a data set derived fromanalysis of the statistical correlationswithina set of alignedtransferRNA sequences. By assigning reasonable physical interpretationsto certainstatistical correlations,weareable to calculate three-dimensionalstructures for tRNAfroma randomstarting structure. Theconstraints that weuse are associatedwithdifferent variances,andsotheir effectsare not uniform,and mustbe reconciled by a probabilistic algorithmto yield the mostlikely structure. Asmightbe predicted,the uncertaintyin the positionfor eachbaseis a functionof boththe numberandstrengthof the constraints, and is reflected in the variances in atomic position calculatedby the algorithm....
Russ B. Altman