AUTOMATED CORRELATION AND AVERAGING OF 3-DIMENSIONAL RECONSTRUCTIONS OBTAINED BY ELECTRON TOMOGRAPHY

We have developed a least-squares refinement procedure that in an auto mated way performs three-dimensional alignment and averaging of object s from multiple reconstructions. The computer implementation aligns th e three-dimensional structures by a two-step procedure that maximizes the density overlap for all objects. First, an initial average density is built by successive incorporation of individual objects, after a g lobal search for their optimal three-dimensional orientations. Second, the initial average is subsequently refined by excluding individual o bjects one at a time, realigning them with the reduced average contain ing all other objects and including them into the average again. The r efinement is repeated until no further change of the average occurs. T he resulting average model is therefore minimally biased by the order in which the individual reconstructions are incorporated into the aver age. The performance of the procedure was tested using a synthetic dat a set of randomly oriented objects with Poisson-distributed noise adde d. The program managed well to align and average the objects at the si gnal/noise ratio 1.0. The increase in signal/noise ratio was in all in vestigated cases almost equal to the expected square root of the numbe r of objects. The program was also successfully tested on a set of aut hentic three-dimensional reconstructions from an in situ specimen cont aining Escherichia coli 70S ribosomes, where the immediate environment of the reconstructed objects may also contain variable amounts of oth er structures. (C) 1997 Academic Press.