A new randomization assay reveals unexpected elements of sequence bias in model 'randomized' gene libraries: implications for biopanning

Although techniques such as biopanning rely heavily upon the screening of r andomized gene libraries, there is surprisingly little information availabl e on the construction of those libraries. In general, it is based on the cl oning of 'randomized' synthetic oligonucleotides, in which given position(s ) contain an equal mixture of all four bases. Yet, many supposedly 'randomi zed' libraries contain significant elements of bias and/or omission. Here, we report the development and validation of a new, PCR-based assay that ena bles rapid examination of library composition both prior to and after cloni ng. By using our assay to analyse model libraries, we demonstrate that the cloning of a given distribution of sequences does not necessarily result in a similarly composed library of clones. Thus, while bias in randomized syn thetic oligonucleotide mixtures can be virtually eliminated by using unequa l ratios of the four phosphoramidites, the use of such mixtures does not en sure retrieval of a truly randomized library. We propose that in the absenc e of a technique to control cloning frequencies, the ability to analyse the composition of libraries after cloning will enhance significantly the qual ity of information derived from those libraries. (C) 2000 Published by Else vier Science B.V. All rights reserved.