PAM Disruption Analysis: Identifying CRISPR-Resistant Variants

Learn how to use supremo_lite to identify genetic variants that disrupt Protospacer Adjacent Motif (PAM) sites, making genomic loci resistant to repeated CRISPR editing.

Example PAM sequences:

  • SpCas9: NGG (e.g., AGG, TGG, CGG, GGG)

  • SaCas9: NNGRRT (more complex pattern)

Setup

import supremo_lite as sl
import numpy as np
from pyfaidx import Fasta
import pandas as pd
import os
import re


print(f"supremo_lite version: {sl.__version__}")

# Load test data
test_data_dir = "../../tests/data"
reference = Fasta(os.path.join(test_data_dir, "test_genome.fa"))

print(f"\nReference genome loaded: {list(reference.keys())}")

# Show chr6 sequence
chr6_seq = reference["chr6"][:80].seq
print(f"\nchr6 sequence:")
print(chr6_seq)
print("                                ^^^")
print("PAM sites for SpCas9 (NGG)")

supremo_lite version: 1.0.0

Reference genome loaded: ['chr1', 'chr2', 'chr3', 'chr4', 'chr5', 'chr6']

chr6 sequence:
AAAGCAAATTCAAATCATCCAAGAATGCCACTTGGAATTTGCGATATTTTTGTTTTTTTTTTTTTAATATTTTACAAAAT
                                ^^^
PAM sites for SpCas9 (NGG)

Basic PAM Disruption: SNV Example

# Create a variant that disrupts the NGG PAM
variant_snv = pd.DataFrame(
    [{"chrom": "chr6", "pos1": 34, "id": ".", "ref": "G", "alt": "T"}]
)

Running PAM Disruption Analysis

Now let’s use get_pam_disrupting_alt_sequences() to identify this PAM-disrupting variant:

# Run PAM disruption analysis
gen_snv = sl.get_pam_disrupting_alt_sequences(
    reference_fn=reference,
    variants_fn=variant_snv,
    seq_len=20,  # 20bp window
    max_pam_distance=10,  # Search within 10bp of variant
    pam_sequence="NGG",  # SpCas9 PAM
    encode=False,  # Get raw strings for visualization
    n_chunks=1,
)

# Unpack the generator to get results
alt_seqs, ref_seqs, metadata = next(gen_snv)

print("PAM Disruption Analysis Results:")
print("=" * 50)
print(f"Number of PAM-disrupting variants: {len(metadata)}")
print(f"\nMetadata (first few columns):")
print(
    metadata[
        [
            "chrom",
            "variant_pos1",
            "ref",
            "alt",
            "pam_ref_sequence",
            "pam_alt_sequence",
            "pam_distance",
        ]
    ].to_string()
)

print(f"\n\nReference sequence (no variant):")
for i, (chrom, start, end, seq) in enumerate(ref_seqs):
    print(f"  {i}: {chrom}:{start}-{end}")
    print(f"      {seq}")

print(f"\nAlternate sequence (variant applied):")
for i, (chrom, start, end, seq) in enumerate(alt_seqs):
    print(f"  {i}: {chrom}:{start}-{end}")
    print(f"      {seq}")

print(f"\n✓ The TGG PAM in reference becomes TTG in alternate (no longer matches NGG)")

PAM Disruption Analysis Results:
==================================================
Number of PAM-disrupting variants: 1

Metadata (first few columns):
  chrom  variant_pos1 ref alt pam_ref_sequence pam_alt_sequence  pam_distance
0  chr6            34   G   T              TGG              TTG             1


Reference sequence (no variant):
  0: chr6:23-43
      AATGCCACTTGGAATTTGCG

Alternate sequence (variant applied):
  0: chr6:23-43
      AATGCCACTTTGAATTTGCG

✓ The TGG PAM in reference becomes TTG in alternate (no longer matches NGG)

Understanding the Output Structure

The function returns a generator that yields tuples for memory-efficient processing:

for alt_seqs, ref_seqs, metadata in get_pam_disrupting_alt_sequences(...):
    # Process each chunk
    pass

# Or get all results at once with n_chunks=1
alt_seqs, ref_seqs, metadata = next(get_pam_disrupting_alt_sequences(..., n_chunks=1))

Return Values

Each yield produces a 3-tuple:

1. alt_seqs: Sequences with variant applied

  • Format when encode=False: List of (chrom, start, end, sequence) tuples

  • Format when encode=True: Stacked array/tensor of shape (n_variants, 4, seq_len)

2. ref_seqs: Reference sequences (no variant)

  • Same format as alt_seqs

  • Use these as baseline for comparison with variant sequences

3. metadata: DataFrame with comprehensive variant and PAM information

Metadata Columns

Standard variant information:

  • chrom: Chromosome name

  • window_start, window_end: Window boundaries (0-based)

  • variant_pos0, variant_pos1: Variant position (0-based and 1-based)

  • ref, alt: Reference and alternate alleles

  • variant_type: Variant classification (SNV, INS, DEL, etc.)

PAM-specific information:

  • pam_site_pos: Position of PAM site within the window (0-based)

  • pam_ref_sequence: PAM sequence in reference (e.g., “TGG”)

  • pam_alt_sequence: PAM sequence after variant (e.g., “TTG”)

  • pam_distance: Distance from variant to PAM start

Next Steps