DaisyBlast

Multi-sample synteny detection via transitive BLAST chaining.

DaisyBlast is a CLI tool for detecting and visualizing synteny blocks (collinear, homologous genomic segments) across multiple FASTA inputs. It performs an all-vs-all BLAST search, “shatters” alignments into non-overlapping windows, and groups them into syntenic blocks using a graph-based approach.

The Problem

BLAST is the gold standard for pairwise nucleotide comparison (A ↔ B), but analyzing multiple samples requires a broader view.

DaisyBlast daisy-chains these isolated hits using a Union-Find graph algorithm to enforce transitivity. If A aligns with B, and B aligns with C, DaisyBlast unifies all three into a single Synteny Group. This enables the visualization of conserved structure across n inputs, moving beyond simple pairwise limitations.

Features

• Automated Pipeline Run BLAST, parse hits, identify synteny groups, and generate plots—all from one command.

• Graph-Based Grouping Union-Find logic chains collinear hits into multi-sample synteny blocks.

• Comprehensive Visualizations

  • Synteny maps: Linear and Circos-style circular plots.
  • Dotplots: Pairwise and combined alignment geometry.
  • Coverage summaries: NCBI-style stacked alignments.

• Robust to Fragmentation Uses a “shattering” algorithm to create clean, non-overlapping windows from complex, overlapping BLAST outputs.

---

Installation

Prerequisites

• Python ≥ 3.8
• NCBI BLAST+ (makeblastdb and blastn must be in your PATH)

Option 1: Install from PyPI (Recommended)

pip install daisyblast

Option 2: Install from source

git clone https://github.com/erinyoung/daisyblast.git
cd daisyblast
pip install .

Verify Installation

daisyblast --help

This should show the help menu.

usage: daisyblast [-h] -i INPUT [INPUT ...] [-o OUTPUT_DIR] [-e EVALUE] [--min_pident MIN_PIDENT] [--min_length MIN_LENGTH] [-n NUM_GROUPS]

DaisyBlast: A tool to find and visualize synteny blocks from a single multi-FASTA file.

options:
  -h, --help            show this help message and exit
  -i INPUT [INPUT ...], --input INPUT [INPUT ...]
                        One or more input FASTA files (e.g., contig1.fa contig2.fa).
  -o OUTPUT_DIR, --output_dir OUTPUT_DIR
                        Directory to save output .bed and .png files. (Default: daisyblast_results)
  -e EVALUE, --evalue EVALUE
                        E-value cutoff for the self-BLAST search. (Default: 1e-10)
  --min_pident MIN_PIDENT
                        Minimum percent identity for a BLAST hit. (Default: 90.0)
  --min_length MIN_LENGTH
                        Minimum alignment length *after* splitting hits. (Default: 200)
  -n NUM_GROUPS, --num_groups NUM_GROUPS
                        Maximum number of groups in final bedfile (Default: 20)

---

Usage

daisyblast -i data/contig1.fasta data/contig2.fasta -o results_dir

Quick Start

You can test the installation using the sample data provided in the repository:

# Run on included test files
daisyblast -i tests/data/test_1.fasta tests/data/test_2.fasta -o test_results

---

Output Overview

1. Synteny Maps (Grouped Blocks)

High-level views of conserved regions. Each color corresponds to a unique Synteny Group shared across sequences.

• Circular Plot: The query sequence forms the outer ring; colored blocks indicate a blast hit shared by two or more input sequences.

  

• Linear Map: Synteny blocks plotted along genomic coordinates.

  

2. Alignment Geometry (Dotplots)

Visualizes raw BLAST hits before grouping. Use these to detect inversions (downward diagonals) or indels (gaps).

• Combined dotplots: All subjects vs. one query on a single plot.

• Pairwise dotplots: One panel per sequence pair.

3. Coverage Summaries

NCBI-style stacked bar charts showing alignment depth and scoring.

• Red/Pink: High-scoring hits (>80 bitscore)
• Blue/Black: Lower scoring hits

4. Data Files

File	Description
final_groups.txt	Final synteny group assignments (Group_ID Sequence Start End)
divided.bed	Shattered genomic windows used in analysis
blast_hits.txt	Raw BLAST format 6 output
trimmed_blast.tsv	BLAST hits trimmed to window boundaries

---

How It Works

Step	Task	Reason
1	Rename	All headers are adjusted to ${filename}__${original_header} to ensure unique IDs and prevent collisions when using multiple input files.
2	BLAST	Performs an all-vs-all blastn search across all input FASTA sequences.
3	Shatter	Parses BLAST hits and breaks genomes into discrete, non-overlapping windows. Why? If Overlap(A,B) and Overlap(B,C) differ in size, shattering creates a common denominator window to allow clean comparison.
4	Trim	Crops each BLAST alignment so it fits strictly within its corresponding shattered window.
5	Group	Applies a Union-Find graph algorithm to chain windows into synteny blocks. Logic: If A ↔ B and B ↔ C, DaisyBlast groups A, B, and C together.
6	Visualize	Generates synteny maps, dotplots, and coverage summaries using matplotlib and pycirclize.

---

AI Attribution

Please note that portions of this codebase were written with the assistance of Google Gemini to accelerate development. The package logo was also AI-generated using Gemini's image creation tools.

Citation

If you use DaisyBlast in your research, please cite:

DaisyBlast: Multi-sample synteny detection via transitive BLAST chaining. GitHub repository: https://github.com/erinyoung/daisyblast

License

Distributed under the MIT License. See LICENSE for more information.