We've developed and open-sourced a deterministic alternative to BLAST/MinHash for DNA identification and comparison. Instead of approximate similarity scores, our method creates an exact "passport" for any genome using SHA-256 window hashing.

How it works:

Genome is split into sliding windows (e.g., 200 bp).

Each window is hashed with SHA-256.

The set of hashes forms a unique passport (collision probability = 0).

Structural similarity between passports is measured via TEES resonance - an FFT-based coherence score (0-1).

Key features:

100% exact identification - identical windows - identical hashes; 1 mutation - different hash.

Deterministic - no heuristics, no random seeds, always the same result.

Irreversible - cannot reconstruct DNA from passport (privacy-friendly).

Fast - comparison of 3 genomes (~500-5000 bp) takes <1 second on a laptop.

Two modes: -_fast (Jaccard only) for screening, --full (TEES resonance) for detailed structural comparison.

Benchmark (hemoglobin HBA1 & INS, window=200 bp, step=100 bp):

Comparison  Identical windows   Jaccard TEES resonance  Kinship index
Human - Mouse (HBA1)    0%  0.000   0.536   0.214
Human - Chicken (HBA1)  0%  0.000   0.540   0.216
Mouse - Chicken (HBA1)  0%  0.000   0.522   0.209

Interpretation: The kinship index (0.21- 0.22) reflects evolutionary distance, while exact hash matching (0%) confirms species-level uniqueness. For identical genomes (e.g., clones), Jaccard would be 1.0.

Potential applications:

Forensic identification - exact match confirmation without alignment.

Pathogen screening - rapid detection of known viral sequences by hash lookup.

Genome auditing - monitoring mutations or insertions over time.

Privacy-preserving databases - store only hashes, not raw sequences.

Repository & code: http://github.com/Dimius0/spectravortex

Direct script: genome_passport_v7.py Documentation: TEES_Genome_Passport_Documentation_EN.md

Question for the community: Would a deterministic, 100% accurate passport system be useful for your work - for example, in rapid pathogen screening, forensic identification, or tracking genomic changes over time? What datasets would you recommend for further benchmarking (e.g., 1000 Genomes, known viral insertion panels)?

Thanks in advance for your feedback!