DRAMS

DRAMS detects and re-aligns mixed-up samples in multi-omics studies by aligning sample information with genotype data across multiple omics platforms to improve data integrity and statistical power.

Key Features:

• Multi-Omics Integration: Requires at least three types of omics data (e.g., genomics, transcriptomics, proteomics) to detect and correct sample mix-ups.
• Logistic Regression Model: Employs a logistic regression model that leverages relationships between multi-omics data to identify potential true sample identities.
• Modified Topological Sorting Algorithm: Uses a modified topological sorting algorithm to systematically re-align samples based on predicted true IDs.
• Genotype-Based Alignment: Aligns sample information with genotype data across multiple omics platforms to determine correct sample assignments.
• Validation on Simulated and Real Data: Applied to both simulated and real datasets for evaluation of detection and re-alignment performance.
• Performance Dependence: Performance improves with an increased number of omics data types and with a reduced proportion of mix-ups.

Scientific Applications:

• Enhancing Statistical Power: Correcting sample misalignments strengthens the statistical power of multi-omics studies and reduces false findings.
• Improving Data Quality: Accurate reassignment of samples enhances overall data quality for large-scale integrative analyses.
• Case Study - PsychENCODE BrainGVEX Project: Applied to BrainGVEX data to detect and correct 201 mix-ups (12.5% of total), correctly reassign all 21 racial identity errors, and increase identified QTL with an average fold increase of 1.62 (FDR < 0.01).

Methodology:

Applies a logistic regression model to relationships among multi-omics data and then a modified topological sorting algorithm to re-align samples; evaluated on simulated and real datasets, with performance improving as the number of omics types increases or the proportion of mix-ups decreases.