Histones Classification Based on EGFET Signals

Abstract

Dysregulation of histones has been implicated in several medical conditions, including various cancers and neurodegenerative disorders. Histone-specific biosensors are key in detecting and quantifying them, advancing our understanding of chromatin dynamics and epigenetic regulation for potential breakthroughs in cancer research and personalized medicine. The focus of this paper is on quantifying a biosensor's ability to distinguish between Human Histones (H4) and non-target analytes. Classification methods are used to provide complementary analysis to biosensor data derived from sensor manufactured using a KU7 RNA aptamer bonded to a gold electrode. The features found provide high classification performance (F1 score over 0.99) and suggest physical insights to the operation of the sensor not provided by typical analysis. Furthermore, machine learning techniques are used in an exploratory analysis to test the effects of faulty manufacturing or differences in testing environments on histone detection accuracy.