C-LSTM SMOTE-RFE & XG BOOST Crispr: Deep Learning Models for Predicting CRISPR/Cas12 Guide RNA Activity

R. Sulakshana¹, Dr. R. Lakshmi²

---

ABSTRACT

The most significant device in the enhancement of human genes is clustered regularly interspaced palindromic repeat (CRISPR) period, which can be directed to any gene target the usage of gRNA and Cas enzyme. Due to the shortcoming of low gRNA activity of the CRISPR structures, it is very much advanced where its activity of gRNA can be predicted. The gRNA activity can be calculated using the frequency of insertion or deletion (indel). In this current study, CNN get optimized through the method of determining by means of different conv layers intensity and clear out kernel length. The study also identifies conventional MLR, CNN and C-SVR, and CNN with LSTM and SMOTE-RFE & XG to enhance the model performance. CNN-LSTM SMOTE-RFE and XG have been applied to predict gRNA activity, and this variant became tested the application of Accuracy, Sensitivity, Specificity, F1-score, Spearman correlation, Root Mean Squared error and Mean Squared Error and it performed excellently. The hybrid version is also better than the ultra-modern version in forecasting gRNA pastime through way of means of up to 45%. Lastly, to prove the hybrid model, the study predicts a frequency of indel in the gRNA sequences used in COVID-19 detection; this could be a useful feature in finding the best gRNA to be used in identification of COVID-19 by CRISPR/Cas12 virus.

Keywords: CRISPR, Indel frequency, COVID-19, Convolutional Neural Network, Xtreme Gradient Boosting, Long Short Term Memory Network, SMOTE-RFE, Guide RNA activity, Deep Learning.

How to cite this article: Sulakshana R, Lakshmi R. C-LSTM SMOTE-RFE & XG BOOST Crispr: Deep Learning Models for Predicting CRISPR/Cas12 Guide RNA Activity. Int J Drug Deliv Technol. 2026;16(6s): 1025-1035; DOI: 10.25258/ijddt.16.6s.134

Source of support: None

Conflict of interest: None