Enhanced and predictive modelling of direct shoot regeneration of Evolvulus alsinoides (L.) using ANN machine learning model and genetic stability studies

Abstract

Several factors interact to regulate direct in vitro shoot regeneration. Optimization of de novo direct regeneration is a primary prerequisite for the success of genetic transformation experiments. However, achieving an optimized protocol is typically difficult due to the high cost and time consumption, as well as the complexity of this process. Hence, the application of computational techniques (machine learning (ML) algorithms) is essential for predicting de novo direct regeneration. This study examined the influence of various concentrations of optimal plant growth regulator (PGR) on the successful de novo in vitro regeneration of the shoot of Evolvulus alsinoides. The de novo direct regeneration of E. alsinoides was modelled using Multilayer Perceptron (MLP). Performance of the model was assessed using computational metrics (RMSE and R²). The outcome demonstrated that the model algorithm had higher predictive accuracy. The mean square error (MSE) value was obtained as 5.18E-02, and the R² value was 0.99565. Moreover, the findings revealed a 90.83 % regeneration rate with 26.25 shoots per explant achieved from Murashige and Skoog (MS) medium supplemented with 2 μM Thidiazuron (TDZ) and Indole-3-acetic acid (IAA) (0.1 μM). Based on our results, the MLP was able to optimize the variables accurately. The results indicated good performance in modelling and optimization of in vitro de novo direct regeneration. The model may be used as a dependable and accurate predictive technique for ensuing investigations in in vitro plant genetic engineering.