orthogonal partial least square (opls) algorithm  (Umetrics)

Journal: Frontiers in Cell and Developmental Biology

Article Title: Epigenetic DNA Methylation Signatures Associated With the Severity of Paget’s Disease of Bone

doi: 10.3389/fcell.2022.903612

Figure Lengend Snippet: Validation of predictive models for PDB severity scores. Orthogonal Partial Least Square (OPLS) classifier was trained on the discover set and tested on the cross-validation set. The features consisted of the list of significant sites from site and region level analysis based on the discovery samples (Panels A and C, n = 2,247 CpG sites) and separately, on a best explanatory subset from this pool of sites obtained with the Glmnet (Panels B and D, n = 33 CpG sites) (A,B) Scatter plots of predicted versus observed PDB_severity scores for each model. The observed PDB severity scores were also dichotomized into two groups based on the median PDB severity score of seven; low severity (PDB_severity_score <7; blue) and high severity (PDB_severity_score ≥7; red) and subsequently used for ROC curve analysis showing the area under curve AUC for pooled sites (C) and Glmnet selected sites (D) .

Article Snippet: The Orthogonal Partial Least Square (OPLS) algorithm by SIMCA ver.15 (Umetrics, Sweden) was used to construct predictive models.

Techniques: Biomarker Discovery