Skip to content

usage-guide.md

Latest commit

 

History

History
60 lines (39 loc) · 1.78 KB

usage-guide.md

File metadata and controls

60 lines (39 loc) · 1.78 KB

Off-Target Prediction - Usage Guide

Overview

Predict and score potential off-target cleavage sites for CRISPR guides using Cas-OFFinder and CFD scoring algorithms.

Prerequisites

# Cas-OFFinder (download binary from http://www.rgenome.net/cas-offinder/)
# Or compile from source:
git clone https://github.com/nickloman/cas-offinder
cd cas-offinder && cmake . && make

pip install pandas biopython

Quick Start

Tell your AI agent what you want to do:

  • "Check off-target sites for my guide sequence ATCGATCGATCGATCGATCG"
  • "Find potential off-targets for these 5 guides in the human genome"
  • "Calculate specificity scores for my CRISPR guides"

Example Prompts

Single Guide Analysis

"Find off-targets for ATCGATCGATCGATCGATCG allowing up to 4 mismatches"

"What is the specificity score for this guide sequence?"

Batch Analysis

"Rank my 10 guides by off-target specificity"

"Find the most specific guide from this list"

Context-Aware

"Check if any off-targets fall in coding regions"

"Flag off-targets in tumor suppressor genes"

What the Agent Will Do

  1. Accept guide sequence(s) and genome reference
  2. Run Cas-OFFinder to identify potential off-target sites
  3. Calculate CFD scores for each off-target
  4. Compute aggregate specificity scores
  5. Flag high-risk off-targets (low mismatch count, high CFD)
  6. Return ranked results with recommendations

Tips

  • Mismatch tolerance - 4 mismatches balances speed and sensitivity; use 3 for faster results
  • CFD threshold - Sites with CFD > 0.5 have significant cleavage risk
  • Exonic off-targets - Prioritize avoiding off-targets in coding regions
  • GPU acceleration - Use GPU mode for faster genome-wide searches
  • Multiple guides - Always compare specificity across candidate guides