SAG-QC: Quality control of single amplified genome information by subtracting non-target sequences based on sequence compositions

Background: Whole genome amplification techniques have enabled the analysis of unexplored genomic information by sequencing of single-amplified genomes (SAGs). Whole genome amplification of single bacteria is currently challenging because contamination often occurs in experimental processes. Thus, to increase the confidence in the analyses of sequenced SAGs, bioinformatics approaches that identify and exclude non-target sequences from SAGs are required. Since currently reported approaches utilize sequence information in public databases, they have limitations when new strains are the targets of interest. Here, we developed a software SAG-QC that identify and exclude non-target sequences independent of database. Results: In our method, "no template control" sequences acquired during WGA were used. We calculated the probability that a sequence was derived from contaminants by comparing k-mer compositions with the no template control sequences. Based on the results of tests using simulated SAG datasets, the accuracy of our method for predicting non-target sequences was higher than that of currently reported techniques. Subsequently, we applied our tool to actual SAG datasets and evaluated the accuracy of the predictions. Conclusions: Our method works independently of public sequence information for distinguishing SAGs from non-target sequences. This method will be effective when employed against SAG sequences of unexplored strains and we anticipate that it will contribute to the correct interpretation of SAGs.