Gel filtration chromatographs of lipoproteins represent a superpositio
n, or convolution, of the intrinsic polydispersity of the solute and t
he dispersion due to transport phenomena. We describe a deconvolution
technique for improving the resolution of gel filtration chromatograph
s applicable to lipoproteins and other polydisperse solutes. A matrix
of spreading functions, characterizing the dispersive properties of th
e column, was determined by fitting chromatographic data from a series
of monodisperse standards with the solution to the transport equation
s and interpolating between the fit parameters. A successive approxima
tion scheme was used in which a test distribution was incrementally co
rrected by an amount proportional to the error between the measured ch
romatograph and that derived from the test distribution. A nonlinear r
elaxing function was used to constrain the correction term such that t
he solution remained physically realizable (i.e., nonnegative absorban
ce) as it evolved. Deconvolved chromatographs of Lipoproteins provided
resolution of peaks that were obscured by spreading in the original d
ata. The distribution of particle sizes within each fraction was calcu
lated and verified experimentally by further separating the contents o
f fractions by gradient gel electrophoresis. Our technique, however, p
rovided comparable resolution of the peaks without the additional expe
rimental procedure. (C) 1996 Academic Press, Inc.