Skip to content

changlabtw/wpSBOOT

BranchesTags

Repository files navigation

wpSBOOT — Weighted Partial Super Bootstrap

License: MIT CI Docker

A phylogenetic support assessment protocol that takes multiple sequence alignments (produced by different alignment tools) as input, builds a weighted super-MSA, and generates bootstrap support values using weighted partial resampling.

Overview

Different alignment tools (ClustalW, MAFFT, Muscle, etc.) produce different alignments from the same input sequences, each with its own biases and uncertainties. wpSBOOT exploits this diversity by:

  1. Weighting each alignment by how much it disagrees with the others — alignments with less inter-aligner agreement carry more unique signal and receive higher weight
  2. Concatenating all alignments into a single super-MSA
  3. Bootstrapping with weighted partial resampling — each replicate draws a fraction (1/N) of sites preferentially from high-weight alignments
  4. Inferring ML and bootstrap trees, then mapping support values

This propagates alignment uncertainty into bootstrap support, providing a more realistic assessment of phylogenetic confidence.

Dependencies

Tool Version Purpose
t_coffee ≥ 13.0 Pairwise alignment similarity
RAxML-NG ≥ 1.0 ML and bootstrap tree inference
Perl ≥ 5.26 Run concatenate.pl
BioPerl ≥ 1.7 Required by concatenate.pl (Bio::AlignIO)
wei_seqboot (included) Weighted partial bootstrap sampling
Python ≥ 3.6 Run support_summary.py (standard library only)

t_coffee and raxml-ng binaries are expected in bin/, or available in PATH. wei_seqboot is compiled from the included C++ source.

Installation

Option A: Docker (easiest — no dependency setup required)

# Pull the pre-built image
docker pull changlabtw/wpsboot

# Run on your data (mount the directory containing your alignments)
docker run --rm -v /path/to/data:/data changlabtw/wpsboot \
    -i /data/aln1.fasta -i /data/aln2.fasta -i /data/aln3.fasta \
    -o /data/results/

# Run the built-in example (output stays inside the container)
docker run --rm changlabtw/wpsboot \
    bash /opt/wpsboot/test.sh

Or build locally from source:

git clone https://github.com/changlabtw/wpSBOOT.git
cd wpSBOOT
docker build -t wpsboot .
docker run --rm -v /path/to/data:/data wpsboot \
    -i /data/aln1.fasta -i /data/aln2.fasta -o /data/results/

HPC users: Docker images are compatible with Singularity/Apptainer without root access:

singularity pull wpsboot.sif docker://changlabtw/wpsboot
singularity run -B /path/to/data:/data wpsboot.sif \
    -i /data/aln1.fasta -i /data/aln2.fasta -o /data/results/

Option B: conda (recommended)

git clone https://github.com/changlabtw/wpSBOOT.git
cd wpSBOOT
conda env create -f environment.yml
conda activate wpsboot
cd src/ && make && cd ..
./test.sh

Coming soon: A conda-lock.yml will be provided for fully reproducible installs (exact package versions pinned). Once available:

conda install -c conda-forge conda-lock
conda-lock install conda-lock.yml
conda activate wpsboot
cd src/ && make && cd ..
./test.sh

Option C: manual

1. Clone the repository

git clone https://github.com/changlabtw/wpSBOOT.git
cd wpSBOOT

2. Compile wei_seqboot

wei_seqboot requires a C++11-compatible compiler (no external libraries needed).

cd src/
make
cd ..
# binary is placed at bin/wei_seqboot

3. Install t_coffee and RAxML-NG

Download binaries and place them in bin/, or install them system-wide so they are in PATH.

4. Install BioPerl

cpan Bio::AlignIO Bio::Align::Utilities Bio::LocatableSeq

5. Verify installation

./test.sh

All checks should pass before running on real data.

Usage

./scripts/wpsboot.sh -i <aln1.fasta> -i <aln2.fasta> [...] -o <output_dir> [options]

Required

Flag Description
-i <file> Input alignment (FASTA). Specify once per alignment — minimum 2 required.
-o <dir> Output directory (created if it does not exist).

Options

Flag Default Description
-n <int> N × 100 Number of bootstrap replicates (N = number of input alignments)
-p <float> 1/N Partial sampling fraction (0.0–1.0)
-m <model> GTR+G Substitution model for RAxML-NG
-T <int> 4 Number of threads
-s <int> — Random seed for reproducible bootstrap sampling (default: time-based)
-f — Force rerun of all steps, ignoring any existing outputs
-k — Keep intermediate per-replicate bootstrap files (default: deleted after step 5)
-h — Show help and exit

Examples

Two yeast gene families are included in example/, each with seven alignments produced by different alignment tools.

Input alignments can be specified individually with -i or with a glob:

# Glob shorthand (requires all files in the directory to be input alignments)
./scripts/wpsboot.sh -i "example/YPL070W/*.fasta" -o results/YPL070W

YPL070W

./scripts/wpsboot.sh \
    -i example/YPL070W/clustalw_YPL070W.fasta \
    -i example/YPL070W/DCA_YPL070W.fasta \
    -i example/YPL070W/dialign_YPL070W.fasta \
    -i example/YPL070W/mafft_YPL070W.fasta \
    -i example/YPL070W/muscle_YPL070W.fasta \
    -i example/YPL070W/probcons_YPL070W.fasta \
    -i example/YPL070W/tcoffee_YPL070W.fasta \
    -o results/YPL070W

YDR192C

./scripts/wpsboot.sh \
    -i example/YDR192C/clustalw_YDR192C.fasta \
    -i example/YDR192C/DCR_YDR192C.fasta \
    -i example/YDR192C/dialign_YDR192C.fasta \
    -i example/YDR192C/mafft_YDR192C.fasta \
    -i example/YDR192C/muscle_YDR192C.fasta \
    -i example/YDR192C/probcons_YDR192C.fasta \
    -i example/YDR192C/tcoffee_YDR192C.fasta \
    -o results/YDR192C

With 7 alignments the defaults resolve to:

  • Bootstrap replicates: 700 (7 × 100)
  • Partial fraction: 0.143 (1/7) — each replicate draws approximately one alignment's worth of sites from the super-MSA

Or run via the test script:

./test.sh                        # quick test, YPL070W (default)
./test.sh --full                 # full test,  YPL070W
./test.sh --gene YDR192C         # quick test, YDR192C
./test.sh --gene YDR192C --full  # full test,  YDR192C
./test.sh --gene all             # quick test, both genes
./test.sh --gene all --full      # full test,  both genes

Output

The final result is written to:

<output_dir>/wpSBOOT_result.nwk    ← ML tree with bootstrap support values
<output_dir>/wpsboot.log           ← full pipeline log

Intermediate files are organised by pipeline step:

<output_dir>/
├── 01_similarity/
│   └── similarity.csv             ← per-alignment avg similarity and weight
├── 02_superMSA/
│   ├── super_aln.phylip           ← concatenated super-MSA (PHYLIP format)
│   └── site_weights.txt           ← one weight per site (input to wei_seqboot)
├── 03_bootstrap/
│   └── outfile                    ← all bootstrap replicates (PHYLIP, concatenated)
├── 04_ml_tree/
│   └── ml_tree.raxml.bestTree     ← ML best tree (Newick)
├── 05_boot_trees/
│   └── bootstrap_trees.nwk        ← all bootstrap trees (one per line)
└── 06_support/
    └── support.raxml.support      ← support tree (also copied to wpSBOOT_result.nwk)

Support summary

A helper script scripts/support_summary.py reports per-node and whole-tree bootstrap support:

# Per-node summary + tree topology
python3 scripts/support_summary.py <output_dir>/wpSBOOT_result.nwk

# Also compute whole-tree topology support
python3 scripts/support_summary.py <output_dir>/wpSBOOT_result.nwk \
    <output_dir>/05_boot_trees/bootstrap_trees.nwk

# Verbose: print each bootstrap tree and its match status
python3 scripts/support_summary.py <output_dir>/wpSBOOT_result.nwk \
    <output_dir>/05_boot_trees/bootstrap_trees.nwk --verbose

Per-node support (from wpSBOOT_result.nwk): the fraction of bootstrap trees containing each bipartition, as mapped by RAxML-NG.

Whole-tree topology support: the fraction of bootstrap trees whose unrooted topology is identical to the ML reference tree (all bipartitions match simultaneously). This is a stricter measure than per-node support — a tree is counted only if every internal branch matches.

Method

Alignment weighting (Step 1)

For each pair of input alignments, t_coffee -other_pg aln_compare -compare_mode column computes a column-wise similarity score. The weight for each alignment is:

weight = 100 − average_pairwise_similarity

Alignments that differ more from the others receive higher weight, reflecting their contribution of unique phylogenetic signal.

Weighted super-MSA (Step 2)

All N input alignments are concatenated in order into a single PHYLIP-format super-MSA using concatenate.pl (BioPerl). Each site in the super-MSA inherits the weight of its source alignment, producing a per-site weight file for wei_seqboot.

Weighted partial bootstrap (Step 3)

wei_seqboot generates bootstrap samples from the super-MSA:

  • Sites are drawn with probability proportional to their weight
  • Only a fraction (default 1/N) of all sites are selected per replicate
  • Selected sites are then resampled with replacement (standard bootstrap)

ML tree inference (Step 4)

raxml-ng infers the maximum-likelihood tree from the full super-MSA using the specified substitution model (default: GTR+G). This serves as the reference topology onto which bootstrap support values are later mapped.

Bootstrap tree inference (Step 5)

raxml-ng infers one ML tree per bootstrap replicate in parallel (up to THREADS jobs at a time). All bootstrap trees are collected into a single file for the final step.

Bootstrap support mapping (Step 6)

raxml-ng --support compares each bootstrap tree against the ML reference tree and annotates each internal branch of the ML tree with the fraction of bootstrap replicates that contain the same bipartition. The annotated tree is written to wpSBOOT_result.nwk.

Repository structure

wpSBOOT/
├── bin/                      ← executables (t_coffee, raxml-ng, wei_seqboot)
├── scripts/
│   ├── wpsboot.sh            ← main pipeline wrapper
│   ├── step1_similarity.sh
│   ├── step2_superMSA.sh
│   ├── step3_bootstrap.sh
│   ├── step4_ml_tree.sh
│   ├── step5_boot_trees.sh
│   ├── step6_support.sh
│   ├── concatenate.pl        ← BioPerl alignment concatenation
│   └── support_summary.py    ← bootstrap support summary (per-node + whole-tree)
├── src/                      ← wei_seqboot C++ source (main.cpp, element.cpp, makefile)
├── example/
│   ├── YPL070W/              ← 7 FASTA alignments for YPL070W
│   └── YDR192C/              ← 7 FASTA alignments for YDR192C
├── test.sh                   ← test script (supports --gene, --full)
└── LICENSE

License

GPL3 License. See LICENSE for details.

Web Server

Note: The web server is currently unavailable. Please use the command-line interface described above.

https://wpsboot.page.link/main

References

About

wpsboot.page.link/main

Topics

bootstrap phylogenetic-trees alignment concatenate

Resources

Readme

License

GPL-3.0 license
Activity
Custom properties

Stars

0 stars

Watchers

0 watching

Forks

1 fork
Report repository

Releases 1

Initial Stable Release Latest
Mar 27, 2026

Packages

 
 
 

Contributors

Languages