Improving the Generation of Viral Consensus Sequences Using Adaptive Models

The increasing amount of diverse human viral sequenced samples has exacerbated the necessity of accurately re-constructing these genomes. This need is particularly pronounced in clinical and forensic scenarios, where the sequenced samples are particularly challenging to reconstruct due to low-depth coverage and high mutation rates. To improve the reconstruction of the sequenced samples, multiple genome sequences belonging to an organism can be combined using a consensus-generating process. In this paper, we describe a novel method, Optimized Adaptive Weighted-K (OAWK). This method can analyse the genome sequences present in an aligned multi-FASTA file and generate a consensus sequence in the FASTA format. This approach is capable of retrieving information from each iteration of the consensus generation process and using it to predict the correctness of the following bases. The results obtained indicate that this methodology can generate accurate consensus sequences, improving the consensus by, on average, 88.5% in terms of the average identity, 46.8% in terms of the NCSD, 55.9% in terms of the NRC and 858.1% in terms of the number reconstructed bases (excluding gaps), in relation to a previously available method. The OAWK is publicly available at https://github.com/cobilab/OAWK.