Cross-Modal Normalization for MaxFuse

概要

This skill provides a specialized workflow for aligning the scales of RNA and protein datasets during cross-modal integration, specifically targeting algorithms like MaxFuse. By enforcing consistent z-scoring and variance normalization across both modalities, it prevents high-variance features (typically RNA) from dominating the matching process. This ensures that distances computed for matching are biologically meaningful, significantly reducing false-positive cell type identifications and improving the accuracy of downstream single-cell analysis.

主な機能

• Optimized workflows for Scanpy and Scikit-learn
• Automated z-scoring for multi-modal datasets
• Outlier clipping and range verification
• Standardized RNA-Protein scale alignment
• Variance mismatch diagnostic checks
• 0 GitHub stars

ユースケース

• Integrating CITE-seq or multi-modal single-cell data
• Correcting cluster-matching errors in MaxFuse integration
• Normalizing protein markers for balanced AI model training