From genomic to panomic predictions: An intuitive PANOMICs platform utilizing advanced Machine Learning algorithms

Advanced sequencing techniques have propelled molecular biology research into what is often termed the post-genomics era. Consequently, there is a heightened focus on unraveling the functional connections between individual genes and their ultimate phenotype, an area referred to as predictive genomics. To support this quest, we developed an intuitive platform, PANOMICS, that we believe substantially facilitates the statistical model building to pathway inference, with applications ranging from agriculture to the biomedical field. PANOMICS, designed using Matlab2023b App Designer, allows for the uploading of genomics data, such as Single Nucleotide Polymorphisms (SNPs) information, metabolomics, other omics data, and phenotype information. The integration of different omics layers for the prediction of phenotypic outcomes promises to improve the accuracy and shifts the original genomic prediction, based solely on genomic inforamtion, to panomic prediction, that integrates multiple layers of molecular organization. PANOMICS facilitates data visualization and predictive analysis using various approaches and methods, such as linear, non-linear, and deep learning approaches. These analyses include statistical prediction models calculated based on Least Absolute Shrinkage and Selection Operator (LASSO), Ridge Regression (RR), Elastic Net Regression (ENR), Gaussian Process Regression (GPR), Support Vector Regression (SVR), Partial Least Squares (PLS) regression, and Random Forest (RF), as well as sophisticated Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) models. Genome-Wide Association Studies (GWAS) with visualization based on an implementation of mixed linear models such as Ridge Regression with Best Linear Unbiased Prediction (rrBLUP) and Polygenic Risk Scores (PRS) are all put into practice within the app. The integration of a full range of advanced techniques within the application enhances its utility for researchers aiming to identify and interpret complex genetic interactions. This can lead to a more efficient use of Machine Learning (ML) algorithms in fields ranging from agriculture to medicine.