Sequence Aligner

Build and tune your own DNA and protein sequence aligner. This repo has two parts:

1. A native Windows app (WinUI 3, C#) where you tune the scoring and watch the alignment change live.
2. A Python engine with the same algorithms, plus a one-click runner that writes all results and charts to a folder.

Both run the same methods: global (Needleman-Wunsch) and local (Smith-Waterman), with a linear or affine gap model, for DNA (match / mismatch) or protein (BLOSUM62 / PAM250).

Download and run (no build needed)

Get the latest Windows build from the Releases page: download SequenceAlignerApp-win-x64.zip, unzip it, and run SequenceAlignerApp.exe. It is self-contained, so you do not need to install .NET or anything else. (The 3D viewer and the AI structure feature need an internet connection.)

The Windows app

Folder: SequenceAlignerApp/

Features:

• Two sequence boxes, load a FASTA file, or pick a built-in sample (toy DNA, hemoglobin, cytochrome c, SARS spike, or whole SARS genomes).
• Sliders for match reward, mismatch penalty, and gap penalty. The alignment redraws as you move them.
• Global, local, or semi-global (free end gaps); DNA or protein; BLOSUM62 or PAM250; linear or affine gaps.
• Handles two whole 30,000-letter genomes with a banded aligner.
• Colored alignment view with a match line, plus score, percent identity, gap counts, and sequence stats.
• Translate DNA to protein, reverse complement, a dot plot, and copy or save the alignment.
• A 3D viewer that predicts your protein's structure (ESMFold) or loads a PDB, and colors it by where the two sequences differ.
• An AI secondary-structure predictor (helix / sheet / coil) that runs on the GPU or NPU with Windows ML.
• An AI protein-analysis panel driven by a transformer language model trained from scratch: it scores how related two proteins are, predicts a protein's family, and finds similar proteins in a built-in database, also on the GPU or NPU.

Build and run (needs the .NET 9 SDK; the Windows App SDK restores automatically):

cd SequenceAlignerApp
dotnet build
dotnet run

The Python engine

• app/engine.py is the alignment engine (no third-party dependencies).
• run.py (or double-click run.bat on Windows) runs every analysis and writes all results and charts into an Output folder.
• app/make_figures.py and app/make_3d.py draw the charts.

Setup:

pip install -r requirements.txt
python run.py

What is in here

Path	What it is
SequenceAlignerApp/	The WinUI 3 native app (C#)
app/engine.py	The Python alignment engine
run.py, run.bat	One-click runner that fills the Output folder
Alignment/	The original algorithm files and the sequence data
figures/	The charts as images
TECHNICAL_DOCUMENTATION.md	Full write-up of every algorithm and model (also as .docx)
SUBMISSION_AND_DEMO_GUIDE.md	What to submit and how to run a live demo (also as .docx)

Documentation

• TECHNICAL_DOCUMENTATION.md explains every algorithm and model in detail: the five aligners (global, local, semi-global, affine, banded), the supporting algorithms and scoring matrices, the GPU batched aligner, the two on-device AI models, and the app. A Word version (TECHNICAL_DOCUMENTATION.docx) is included for reading offline.
• SUBMISSION_AND_DEMO_GUIDE.md lists exactly what to hand in and gives a rehearsed, timed live-demo script for both the notebook and the app.
• A shorter plain-English overview is in Alignment/DOCUMENTATION.md.

The science, in one line

Changing the scoring changes the alignment, so the score alone cannot tell you which alignment is right. Percent identity, sensible gaps, and biology have to.