A method for automated identification by sequence and quantitative evaluation of expressed genes, which is unaffected by sequence-dependent variations in hybridization efficiencies. A microbead cDNA library, one DNA fragment per bead, is immobilised in a planar flow cell that allows exposure to a sequence of reagents and scanning of the cell to record the positions of attached fluorescent tags. The entire library is sequenced simultaneously in cycles; during the cycle, each bead reports a four-base sequence, so four or five cycles are sufficient to unambiguously identify a fragment. To initiate the procedure, an endonuclease exposes four-base overhangs. The sequences of these overhangs are identified in four steps to identify the first, second, third and fourth base by ligation to a characteristically tagged adaptor oligonucleotide. In the first step, 16 adaptors that are presented have the first (5′) nucleotide paired with a 3′ sequence that is designed to accept one of four fluorescent tags; the next three 5′ nucleotides are randomised to allow all possible template sequences to bind one of the set of 16 oligonucleotides. The positions of the fluorophores within the flow cell are recorded and the adaptors cleaved. In the next three steps, the second, third and fourth positions are similarly determined. The cycle ends with the exposed 5′ ends being shortened to allow the next four-base sequence to be determined in the next cycle. Because of the numbers of beads that are accommodated by this method, each cDNA may be represented up to hundreds of times; the number of beads that contain each unique sequence, therefore, is a measure of gene expression.