babappalign 1.1.3

Embedding-first deep learning multiple sequence alignment engine with affine-gap DP

BABAPPAlign

Overview

BABAPPAlign is an embedding-first progressive multiple sequence alignment (MSA) engine for protein sequences. It integrates pretrained protein language model embeddings with a learned neural residue–residue scoring function within a classical, exact affine-gap dynamic programming framework (Gotoh).

The method is designed to improve alignment accuracy while maintaining methodological transparency and full reproducibility. BABAPPAlign is fully functional on CPU-only systems; GPU acceleration is optional and affects performance only, not correctness.

---

Key features

• Progressive multiple sequence alignment (MSA)
• Strict learned residue–residue scoring model (BABAPPAScore)
• Uses pretrained protein language model residue embeddings
• Column-aware profile scoring
• True affine-gap dynamic programming (Gotoh algorithm)
• Exact dynamic programming (no heuristics inside DP)
• Embedding inference performed outside DP
• Fully functional on CPU-only systems
• Optional GPU acceleration for faster embedding and scoring
• Explicit model specification (no silent fallback)
• Reproducible and Bioconda-compliant design

---

Installation

Install from PyPI

pip install babappalign

Install from Bioconda

conda install -c bioconda babappalign

This installs a CPU-compatible version of BABAPPAlign. No GPU, CUDA, or special hardware is required.

---

Quick start

babappalign input.fasta -o output.aln.fasta --model babappascore

Important:
BABAPPAlign requires an external trained neural scoring model. The model is not downloaded automatically and must be obtained explicitly (see below).

---

How BABAPPAlign works

1. Residue embedding
   Each protein sequence is converted into residue-level embeddings using a pretrained protein language model.

2. Learned residue scoring
   Residue compatibility is evaluated using a pretrained neural scoring model (BABAPPAScore), replacing traditional substitution matrices.

3. Progressive alignment
   Sequences are progressively aligned using exact affine-gap dynamic programming (Gotoh). Neural inference is performed outside the DP recursion to preserve correctness.

The progressive ordering is a computational heuristic and is not interpreted as a phylogeny.

---

Model weights (required)

BABAPPAlign requires a trained neural residue-level scoring model (BABAPPAScore), which is distributed separately via Zenodo.

Concept DOI (all versions):
https://doi.org/10.5281/zenodo.18053200

Version-specific DOIs are provided on Zenodo for exact reproducibility.

Download and use

# 1. Download the model (one-time)
mkdir -p ~/.cache/babappalign/models

wget https://zenodo.org/record/18053201/files/babappascore.pt      -O ~/.cache/babappalign/models/babappascore.pt

# 2a. Run BABAPPAlign using the cached model name (recommended)
babappalign input.fasta -o aligned.fasta --model babappascore

# 2b. OR run BABAPPAlign using an explicit model path (equivalent)
babappalign input.fasta -o aligned.fasta \
  --model ~/.cache/babappalign/models/babappascore.pt

At runtime, BABAPPAlign prints the resolved model path and a SHA-256 checksum to ensure transparent and reproducible model usage.

---

CPU and GPU execution

BABAPPAlign produces identical alignments on CPU and GPU. GPU acceleration affects performance only.

Component	CPU	GPU
Progressive alignment (DP)	Yes	Yes
Learned scoring	Yes	Yes
Embedding generation	Slower	Faster

---

Input requirements

• Protein sequences only
• FASTA format
• No strict limits on sequence length or number (runtime depends on hardware)

---

Command-line interface

babappalign --help

Key options include:

• -o, --output FILE : output alignment file
• --model MODEL : scoring model name or path (mandatory)
• --gap-open FLOAT : gap opening penalty
• --gap-extend FLOAT : gap extension penalty
• --device {cpu,cuda} : select execution device

---

License

MIT License. See the LICENSE file for details.

---

Citation

Manuscript in preparation.

---

Author and repository

• Author: Krishnendu Sinha
• GitHub: https://github.com/sinhakrishnendu/BABAPPAlign