Towards a more accurate error model for BioNano optical maps

Abstract

Next-generation sequencing technologies has advanced our knowledge in genomics by a tremendous step in the past years. On the other hand, there are still critical questions left unanswered due to the intrinsic limitations of short read length. To address this issue, several new sequencing platforms came into view. However, a lack of comprehensive understanding of the sequencing error poses a primary challenge for their optimal use. Here, we focus on optical mapping, a high-throughput laboratory technique that provides long-range information of a genome. Existing error model is based on OpGen maps. It is not clear if the model is also good for BioNano maps. In this paper, we try to provide a more accurate error model for BioNano optical maps based on real data. Due to the limited availability of real datasets, as an indirect validation of our model, we predict the regions that are difficult to cover and compare the predicted results with the empirical results (both simulated and real data) on human chromosomes. The results are promising, with most of the difficult regions correctly predicted. Tested on BioNano maps, our model is more accurate than the most popular existing error model developed based on OpGen maps. Although we may not have captured all possible errors of the technology, our model should provide important insights for the development of downstream analysis tools using BioNano optical maps.