hairloom 0.0.1

A package to analyze split reads of long read data

Hairloom

Hairloom is a Python package for analyzing split-read alignments from long-read sequencing data. It provides tools to extract read-level split-read tables from BAM files and process these alignments into breakpoint, segment, and translocation tables for downstream analysis.

Features

• Extract and normalize split-read alignments from BAM files.
• Generate tables for:
  • Breakpoints
  • Genomic Segments
  • Translocations
• Designed for compatibility with long-read sequencing data from platforms like ONT and PacBio.
• Simple integration with downstream genomic workflows.

---

Installation

pip install hairloom

---

Usage

Below is an example workflow using Hairloom:

Input

• A BAM file (example.bam).
• A specific genomic region (chrom, start, end).

Step 1: Extract Read-Level Split Read Table

import pysam
from hairloom import extract_read_data

# Open the BAM file
bam_path = "tests/data/test_reads.bam"
bam_file = pysam.AlignmentFile(bam_path, "rb")

# Extract split-read alignment table for a region
chrom, start, end = "chr1", 50, 150
read_table = extract_read_data(bam_file, chrom, start, end)

print("Extracted Read-Level Table:")
print(read_table)

Output: A table of split-read alignments with columns like:

• qname: Read name.
• chrom: Chromosome.
• start: Alignment start position.
• end: Alignment end position.
• strand: Strand information.
• clip1, clip2: Soft/hard clip lengths.
• match: Number of matched bases.
• pclip1: Strand-corrected clip length.

Extracted Read-Level Split-Read Table:
   qname chrom  start   end strand  clip1  match  clip2  pclip1
0  read1  chr1    101   201      +      0    100    300       0
1  read1  chr2    300   400      +    100    100    200     100
2  read1  chr2    700   800      -    100    100    200     200
3  read1  chr2    900  1000      +    300    100      0     300

---

Step 2: Generate Breakpoint Table

from hairloom import make_bundle, make_brk_table, make_brk_supports

# Make bundled BreakpointChain from read table
bundle = make_bundle(read_table)

# Create a breakpoint table
breakpoint_table = make_brk_table(bundle)

print(breakpoint_table)

Output: A table of breakpoints with columns:

• chrom: Chromosome.
• pos: Position.
• ori: Orientation (+ or -).
• support: Support count.

  chrom  pos ori  support
0  chr1  201   +        1
1  chr2  300   -        1
2  chr2  400   +        1
3  chr2  700   -        1
4  chr2  800   +        1
5  chr2  900   -        1

---

Step 3: Generate Segment Table

from hairloom import make_seg_table

# Generate segment table
segment_table = make_seg_table(bundle)

print(segment_table.head())

Output: A table of genomic segments with columns:

• chrom: Chromosome.
• pos1: Start position.
• pos2: End position.
• support: Support count.

  chrom  pos1  pos2  support
0  chr2   300   400        1
1  chr2   700   800        1

---

Step 4: Generate Translocation Table

from hairloom import make_tra_table

# Create a translocation table
translocation_table = make_tra_table(bundle)

print(translocation_table.head())

Output: A table of translocations with columns:

• chrom1, pos1, ori1: First breakpoint information.
• chrom2, pos2, ori2: Second breakpoint information.
• support: Support count.

  chrom1  pos1 ori1 chrom2  pos2 ori2  support
0   chr1   201    +   chr2   300    -        1
1   chr2   400    +   chr2   800    +        1
2   chr2   700    -   chr2   900    -        1

---

Examples

Full Workflow

import pysam
from hairloom import (
    extract_read_data,
    make_brk_table,
    make_seg_table,
    make_tra_table,
)

# Open BAM file
bam_file = pysam.AlignmentFile("tests/data/test_reads.bam", "rb")

# Step 1: Extract read-level split-read table
chrom, start, end = "chr1", 50, 150
read_table = extract_read_data(bam_file, chrom, start, end)
bundle = make_bundle(read_table)

# Step 2: Generate breakpoint table
breakpoint_table = make_brk_table(bundle)

# Step 3: Generate segment table
segment_table = make_seg_table(bundle)

# Step 4: Generate translocation table
translocation_table = make_tra_table(bundle)

# Print results
print("Read Table:")
print(read_table.head())
print("\nBreakpoint Table:")
print(breakpoint_table.head())
print("\nSegment Table:")
print(segment_table.head())
print("\nTranslocation Table:")
print(translocation_table.head())

Normalize an SV Table

The normalize_sv_table function sorts and normalizes the breakpoint pairs in a structural variant (SV) table, ensuring consistent ordering for downstream analyses.

import pandas as pd
from hairloom import normalize_sv_table

# Example SV table
sv_data = {
    "chromosome_1": ["chr1", "chr2", "chr1", "chr3"],
    "position_1": [200, 500, 300, 100],
    "strand_1": ["+", "-", "+", "-"],
    "chromosome_2": ["chr1", "chr2", "chr1", "chr2"],
    "position_2": [100, 400, 400, 200],
    "strand_2": ["-", "+", "-", "+"]
}
sv_table = pd.DataFrame(sv_data)

# Normalize the SV table
normalized_sv = normalize_sv_table(sv_table)

print("Original SV Table:")
print(sv_table)
print("\nNormalized SV Table:")
print(normalized_sv)

Output:

Original SV Table:
  chromosome_1  position_1 strand_1 chromosome_2  position_2 strand_2
0         chr1         200        +         chr1         100       -
1         chr2         500        -         chr2         400       +
2         chr1         300        +         chr1         400       -
3         chr3         100        -         chr2         200       +

Normalized SV Table:
  chromosome_1  position_1 strand_1 chromosome_2  position_2 strand_2
0         chr1         100        -         chr1         200       +
1         chr2         400        +         chr2         500       -
2         chr1         300        +         chr1         400       -
3         chr2         200        +         chr3         100       -

Get Structural Variant Type

The get_svtype function determines the type of structural variant (SV) represented by a BreakpointPair. It supports common SV types: translocation (TRA), inversion (INV), duplication (DUP), and deletion (DEL).

from hairloom import Breakpoint, BreakpointPair, get_svtype

# Example BreakpointPairs
tra1 = BreakpointPair(Breakpoint("chr1", 100, "+"), Breakpoint("chr2", 200, "-"))
inv = BreakpointPair(Breakpoint("chr1", 100, "+"), Breakpoint("chr1", 200, "+"))
del_sv = BreakpointPair(Breakpoint("chr1", 300, "+"), Breakpoint("chr1", 500, "-"))
dup = BreakpointPair(Breakpoint("chr1", 500, "-"), Breakpoint("chr1", 700, "+"))

# Get SV types
print("SV Types:")
print(f"TRA: {get_svtype(tra1)}")  # Output: TRA
print(f"INV: {get_svtype(inv)}")    # Output: INV
print(f"DEL: {get_svtype(del_sv)}")  # Output: DEL
print(f"DUP: {get_svtype(dup)}")    # Output: DUP

Output:

SV Types:
TRA: TRA
INV: INV
DEL: DEL
DUP: DUP

---

Contributing

Contributions are welcome! Please submit issues and pull requests via the GitHub repository.

---

License

This project is licensed under the MIT License. See the LICENSE file for details.

---

Feel free to adapt this README.md to fit your exact requirements or additional features in the Hairloom package!