Genome-wide Machine Learning Analysis of Anosmia and Ageusia with COVID-19

Area of Science:

• Genetics
• Infectious Diseases
• Neuroscience

Background:

• The COVID-19 pandemic has led to widespread health and societal disruptions.
• Loss of smell (anosmia) and taste (ageusia) are common, sometimes prolonged, symptoms of COVID-19.
• Understanding the genetic basis of these sensory impairments is crucial for developing targeted treatments.

Purpose of the Study:

• To investigate the host genetic factors contributing to anosmia and/or ageusia in COVID-19 patients.
• To develop and apply a novel machine learning pipeline for variant and gene prioritization.
• To identify candidate genes and variants associated with COVID-19-induced sensory loss.

Main Methods:

• Utilized human genome sequencing data from the University of Iowa (UI) and All of Us (AoU) Research Program COVID-19 cohorts.
• Employed feature selection techniques and machine learning for patient classification and variant/gene prioritization.
• Assessed model performance using permutation-based variable importance (PVI) and area under the receiver operating characteristic (AUROC) curves.

Main Results:

• Achieved high AUROC scores for variant and gene analysis in the UI cohort (up to 0.798) and variant analysis in the AoU cohort (0.687).
• Prioritized several novel and known genetic factors implicated in immune, neuronal, and calcium signaling pathways.
• Identified potential host genetic contributors to COVID-19-related anosmia and ageusia.

Conclusions:

• The study successfully identified candidate host genetic factors associated with COVID-19-related sensory impairments.
• Findings support existing hypotheses regarding the role of immune and neuronal signaling in anosmia/ageusia.
• This research provides a foundation for understanding the sensorineural aspects of COVID-19 and developing future therapeutic strategies.