---
name: crispr-sgrna-predictor
description: Predict CRISPR sgRNA efficiency using Doench Rules and ensemble scoring, assess off-target risk from sequence motifs, and optionally validate Cas-gRNA-DNA complex structures with AlphaFold 3.
allowed-tools: WebFetch, Bash(python *), Bash(mkdir *), Bash(cp *), Bash(ls *), Bash(jq *), Bash(cd *)
---

# CRISPR sgRNA Efficiency & Complex Structure Predictor

## Purpose

This skill provides a comprehensive computational pipeline for CRISPR guide RNA (sgRNA) design, combining:

1. **sgRNA efficiency prediction** using ensemble machine learning features
2. **Off-target risk assessment** based on sequence motif analysis
3. **Optional AlphaFold 3 structural validation** for Cas-gRNA-DNA ternary complexes

## Scientific Background

### CRISPR-Cas9 Mechanism

CRISPR-Cas9 is an RNA-guided endonuclease that induces double-strand breaks (DSBs) at genomic loci complementary to a guide RNA sequence. The sgRNA consists of:
- **Spacer**: 20 nucleotide sequence that binds target DNA
- **Scaffold**: Constant region forming the Cas9-binding structure

### Key Factors Affecting sgRNA Efficiency

| Factor | Impact | Optimal Range |
|--------|--------|---------------|
| GC Content | Secondary structure stability | 40-70% |
| Spacer Length | R-loop formation | 20nt (SpCas9) |
| Position 1 | Doench Rules: C/G preferred | C or G |
| Position 20 | Doench Rules: A/T preferred | A or T |
| Self-Complementarity | Seed region folding | <50% |
| Poly-T Tracts | Pol III termination | Avoid ?? consecutive T |
| PAM Proximity | Cas9 binding initiation | N/A |

## Input Specification

### Required Input Format

```json
{
  "sequence": "GCCAACTTCACCAAGGCCAGTG",
  "target": "GCCAACTTCACCAAGGCCAG",
  "pam": "NGG",
  "cas_variant": "SpCas9"
}
```

### Input Parameters

| Parameter | Type | Required | Default | Description |
|-----------|------|----------|---------|-------------|
| `sequence` | string | Yes | - | Full sgRNA sequence (20-23nt) |
| `target` | string | Yes | - | Target genomic sequence (20nt) |
| `pam` | string | Yes | NGG | PAM sequence (SpCas9: NGG) |
| `cas_variant` | string | No | SpCas9 | Cas protein variant |

### Supported Cas Variants

| Variant | PAM | Spacer Length | Notes |
|---------|-----|---------------|-------|
| SpCas9 | NGG | 20nt | Standard, most common |
| eSpCas9 | NGG | 20nt | Enhanced specificity |
| SpCas9-HF1 | NGG | 20nt | High-fidelity variant |
| SaCas9 | NNGRRT | 21nt | SmallerCas protein |
| AsCas12a | TTTN | 23nt | Different PAM, sticky ends |
| LbCas12a | TTTV | 20nt | Different PAM, sticky ends |
| CasRx | NGG | 22nt | RNA targeting |

## Algorithm Specification

### Scoring Feature Weights

The efficiency score is calculated as a weighted ensemble:

```
Efficiency = 0.15 ? GC_score +
             0.20 ? Positional_score +
             0.15 ? Thermodynamic_score +
             0.15 ? SelfComplementarity_score +
             0.15 ? Pattern_score +
             0.10 ? Length_score
```

### Feature Details

#### 1. GC Content Score (Weight: 15%)

GC content affects DNA melting temperature and secondary structure.

| GC Range | Score | Interpretation |
|----------|-------|----------------|
| 40-60% | 1.0 | Optimal |
| 30-40% or 60-70% | 0.7 | Acceptable |
| <30% or >70% | 0.3 | Suboptimal |

#### 2. Positional Score (Weight: 20%)

Based on Doench Rules (Nature Biotechnology 2014, 2016):

**Position-specific nucleotide weights for SpCas9:**

| Position | Preferred | Avoided | Weight |
|----------|-----------|---------|--------|
| 1 | G, C | A, T | ?0.3-0.5 |
| 2 | G, C | T | ?0.1 |
| 3 | G | A, T | ?0.2 |
| 20 (PAM-adjacent) | A, T | G, C | ?0.2-0.3 |

#### 3. Thermodynamic Score (Weight: 15%)

Nearest-neighbor DNA stability model (SantaLucia 1998):

| Dinucleotide | ?G (kcal/mol) |
|--------------|----------------|
| CG | -2.17 |
| GC | -2.24 |
| GG/CC | -1.97 |
| CA/TG | -1.45 |

Lower ?G (more negative) indicates higher stability. Optimal range: -15 to -25 kcal/mol.

#### 4. Self-Complementarity Score (Weight: 15%)

Measures potential internal base pairing:

```
SelfComp_score = 1.0 - (matches / max_possible) / 2
```

- Compare sequence with its reverse complement
- Count complementary base pairs
- Higher complementarity = lower score

#### 5. Pattern Score (Weight: 15%)

Penalizes harmful sequence motifs:

| Pattern | Penalty | Severity |
|---------|---------|----------|
| Poly-T ?? | -3.0 | High (Pol III termination) |
| Poly-A ?? | -2.0 | High |
| Poly-C ?? | -1.0 | Medium |
| Poly-G ?? | -1.0 | Medium |
| Poly-AT ?? | -2.0 | High |
| 4+ consecutive same | -1.0 | Low |

#### 6. Length Score (Weight: 10%)

| Length | Score |
|--------|-------|
| 20nt | 1.0 |
| 19nt | 0.8 |
| 21nt | 0.85 |
| 22nt | 0.7 |
| 23nt | 0.6 |

## Output Specification

### JSON Output Format

```json
{
  "status": "success",
  "input": {
    "sequence": "GCCAACTTCACCAAGGCCAGTG",
    "target": "GCCAACTTCACCAAGGCCAG",
    "pam": "NGG",
    "cas_variant": "SpCas9"
  },
  "prediction": {
    "efficiency_score": 80.27,
    "efficiency_rank": "High",
    "confidence": "Medium",
    "gc_content": 59.1,
    "gc_content_optimal": true,
    "self_complementarity": 15.0,
    "thermodynamic_score": -18.5
  },
  "off_target_assessment": {
    "risk_level": "Low",
    "risk_score": 1,
    "risk_factors": []
  },
  "sequence_analysis": {
    "length": 22,
    "gc_count": {"G": 7, "C": 6, "A": 5, "T": 4},
    "motifs_found": [],
    "warnings": []
  },
  "recommendations": [
    "GC content is optimal (59.1%)",
    "No unfavorable sequence patterns detected",
    "Self-complementarity is within acceptable range"
  ]
}
```

### Score Interpretation

| Efficiency Score | Rank | Recommendation |
|------------------|------|----------------|
| ??0 | Excellent | Strongly recommended |
| 70-79 | High | Recommended |
| 60-69 | Medium | Acceptable, validate experimentally |
| 50-59 | Low | Consider alternatives |
| <50 | Poor | Not recommended |

### Off-Target Risk Levels

| Risk Level | Score Range | Action |
|------------|-------------|--------|
| Low | 0-1 | Proceed with design |
| Medium | 2-3 | Validate with off-target prediction tools |
| High | ?? | Redesign sgRNA |

## Usage Examples

### Basic Usage

```bash
python execute.py --sequence GCCAACTTCACCAAGGCCAGTG \
                  --target GCCAACTTCACCAAGGCCAG \
                  --pam NGG \
                  --cas SpCas9
```

### Full Output with Report

```bash
python execute.py --sequence GCCAACTTCACCAAGGCCAGTG \
                  --target GCCAACTTCACCAAGGCCAG \
                  --pam NGG \
                  --cas SpCas9 \
                  --output results/sgrna_analysis.json \
                  --report results/sgrna_report.md
```

### Batch Processing

```bash
# Process multiple sequences from JSON file
python execute.py --batch sequences.json --output-dir results/
```

## Limitations & Caveats

### Computational Limitations

1. **Sequence-based only**: Does not perform genome-wide off-target search
2. **SpCas9-centric**: Optimized for standard SpCas9, other variants may have reduced accuracy
3. **Epigenetic factors ignored**: Chromatin accessibility, DNA methylation not considered
4. **Species-specific effects**: Training data may bias toward human/mouse

### Recommendations for Experimental Validation

1. **Off-target sequencing**: Perform GUIDE-seq or CIRCLE-seq for comprehensive off-target detection
2. **Multiple sgRNAs**: Design 3-5 independent sgRNAs per target
3. **Empirical testing**: Validate top candidates in cell-based assays
4. **Seed region conservation**: Consider target site evolutionary conservation

## AlphaFold 3 Integration

### Purpose

Optional structural validation of Cas9-sgRNA-DNA ternary complex.

### Use Cases

1. **PAM recognition validation**: Verify Cas9-PAM-DNA interactions
2. **R-loop formation**: Analyze strand invasion mechanics
3. **Domain positioning**: Check Cas9 conformational changes

### Command

```bash
# Requires AlphaFold 3 server access
python execute.py --sequence <sgRNA> \
                 --alphafold3 \
                 --output-complex complex.pdb
```

### Output

- PDB file with predicted complex structure
- Confidence metrics (pLDDT, PAE)
- Interface analysis between Cas9, sgRNA, and DNA

## Installation & Requirements

### Prerequisites

- Python 3.8+
- Biopython >= 1.79
- NumPy >= 1.20

### Installation

```bash
pip install biopython numpy
```

## References

1. **Doench et al. (2014)**: Rational design of highly active sgRNAs. *Nature Biotechnology* 32:1262-1267
   - doi:10.1038/nbt.3026

2. **Doench et al. (2016)**: Optimized sgRNA design. *Nature Biotechnology* 34:184-191
   - doi:10.1038/nbt.3437

3. **Chuai et al. (2018)**: DeepCRISPR for sgRNA design. *Genome Biology* 19:80
   - doi:10.1186/s13059-018-1459-4

4. **Klein et al. (2025)**: GuideScan2 for gRNA design. *Genome Biology*

5. **Abramson et al. (2024)**: AlphaFold 3 for biomolecular structures. *Nature*
   - doi:10.1038/s41586-024-07487-w

6. **SantaLucia (1998)**: Unified DNA nearest-neighbor thermodynamics. *PNAS* 95:1460-1465

## Author

jsy

## Version History

| Version | Date | Changes |
|---------|------|---------|
| 1.0 | 2026-04-29 | Initial release with efficiency prediction and off-target assessment |
| 1.1 | 2026-04-29 | Added AlphaFold 3 integration, improved scoring model, detailed feature breakdown |