Abstract:One of the practical difficulties that remains in large-scale DNA fragment assembly is the correct reconstruction of DNA sequences including repeats. An approach based on the definite-sized characteristic substring for the masking-off of repeats is proposed after considering the relative position information contained in fragment data. Before pair-wise alignment the approach chose unique substrings to mark fragments for the sake of decrease in possible incorrect overlaps. We also concretely describes the determination of some parameters and finally presents the computational result to prove the effectiveness of the method.