Wavelet-based cancer drug recommender system

Abstract Molecular nature of cancer is the foundation systematic studies genomes, providing exceptional insights and allowing treatments advancement in clinic. We combine techniques image processing for feature enhancement recommender systems proposing a personalized ranking drugs. use database containing drug sensitivity data more than 310.000 IC50, describing response 300 anticancer drugs across 987 cell lines. The system implemented Python (Google Colaboratory) succeed to find best fitted After several preprocessing tasks, regarding data, two experiments are performed. First experiment uses original DNA microarray images second one wavelet transforms preprocess images. Our main goal assess impact using transformed (versus images) on proposed framework. show that, by improving search lines with similar profile new line, produce better results, not only terms evaluation metrics (hit-rate average reciprocal hit-rate), but also execution time.