Intelligent predictor using cancer-related biologically information extraction from cancer transcriptomes
Tahira Shehzadi, Abdul Majid, Madiha Hameed, Aqeel Farooq, Aqsa Yousaf
Abstract
This research includes cancer transcriptome which comprises of modifications in envoi RNAs while cancer genome involves DNA based alterations such as gene duplication and point mutations. Collectively, genome and transcriptome provide an overall view of individual patient's cancer that would impact on clinical decision making. This development has motivated to develop ML-based intelligent models for cancer detection. In this paper, we built predictor to extract cancer-related information using RNA sequences. Here, in this work, an effective RNA bio-sequencing is used and a novel feature extraction technique is applied accordingly. The features are trained on different learning models namely XGBoost, SVM, AdaBoost, Bagging, Random Forest, Naive Bayes, and Linear Regression to extract cancer-related biological information from cancer transcriptomic data. The XGBoost classifier gives the best performance as it provides accuracy 91.67% and sensitivity 96.89% on a standard dataset and independent dataset. These models are useful for precision medicine, drug discovery and clinical oncology.