GenoScene Processed SNP Dosage Matrix Based on the HIrisPlex-S Panel for Forensic DNA Phenotyping

This dataset contains the processed SNP dosage matrix used in the GenoScene forensic DNA phenotyping study. It comprises 10,317 samples and 40 HIrisPlex-S-derived SNP dosage variables (encoded as 0, 1, or 2), along with phenotype labels for eye color, hair color, and skin tone. Data sources include openSNP, the 1000 Genomes Project (Phase 3, GRCh37), a Kazakhstan genomic dataset, and a Caribbean population dataset. The dataset was preprocessed using bcftools and PLINK in a Linux environment. One variant (rs312262906) was excluded due to its identification as an indel. Missing genotypes were imputed using the 1000 Genomes Phase 3 reference panel with an imputation quality threshold of R² > 0.3. The uploaded file represents the final processed dataset used for training and evaluating machine learning models for pigmentation prediction. This dataset supports the manuscript entitled: "GenoScene: AI-Powered Forensic DNA Phenotyping for Pigmentation Prediction and Phenotype-Guided Facial Visualization." The dataset is provided to support transparency, reproducibility, and future research in forensic DNA phenotyping. No personally identifiable information is included.