Fig. 2

The process of construction of the predictive model. a The building of the predictive model composed of data preparation, feature calculation, feature selection and machine learning model training. (Key parameters are in parentheses.) b Schematic diagram of generating a stable range of clustered sequences in a cluster. The sequences in the cluster were aligned against the assembled sequence. The probability of the major nucleotide at each position was computed. A threshold of 0.8 was selected to determine the stable range of the cluster sequence