FANCY

FANCY estimates the number of Single Nucleotide Variants (SNVs) that may leak from raw functional genomics sequencing reads to quantify privacy risk without requiring explicit genotyping.

Key Features:

• Data compatibility: Supports RNA-Seq, ATAC-Seq, and ChIP-Seq read data for privacy risk assessment.
• Privacy risk estimation: Uses a supervised regression model that leverages overall sequencing statistics as features to predict the cumulative number of SNVs likely to leak.
• Predictive performance: Demonstrates an average R² of 0.95 across independent test sets and includes a specialized model variant to improve accuracy when the number of leaking variants is low.
• Implementation and computational resources: Implemented in Python and MATLAB with a GitHub repository (https://github.com/gersteinlab/FANCY) providing custom scripts for feature generation and Jupyter notebooks for parameter optimization.

Scientific Applications:

• Privacy assessment in functional genomics: Provides quantitative estimates of variant leakage risk from RNA-Seq, ATAC-Seq, and ChIP-Seq datasets.
• Data release decision support: Enables prediction of potential privacy leakage to inform data sharing and release policies.
• Participant confidentiality evaluation: Facilitates evaluation of risks to participant confidentiality by estimating likely leaked SNVs.

Methodology:

Applies a supervised regression model using overall sequencing statistics as features to predict the cumulative number of SNVs likely to leak, with feature generation via custom scripts and parameter optimization via provided Jupyter notebooks.