If you have questions about your rights to use, distribute this software, or use for commercial purposes, please contact Berkeley Lab's Intellectual Property Office at IPO@lbl.gov.
# # Test the conversion
message = """
🌐 SpectraCodec: Hilbert curve metadata encoding 🌐
العربية: سبيكترا كودك: ترميز البيانات بمنحنى هيلبرت
हिंदी: स्पेक्ट्राकोडेक: हिल्बर्ट वक्र डेटा एन्कोडिंग
ქართული: სპექტრაკოდეკი: ჰილბერტის მრუდი
ไทย: สเปกตราโคเดก: การเข้ารหัสข้อมูลด้วยเส้นโค้งฮิลเบิร์ต
日本語: スペクトラコーデック：ヒルベルト曲線符号化
አማርኛ: ስፔክትራኮዴክ፡ ሂልበርት ኩርቭ ኢንኮዲንግ
தமிழ்: ஸ்பெக்ட்ராகோடெக்: ஹில்பர்ட் வளைவு குறியீடு
Монгол: СпектраКодек: Хилбертийн муруй
Ελληνικά: ΣπέκτραΚόντεκ: Κωδικοποίηση καμπύλης Χίλμπερτ
עברית: ספקטראקודק: קידוד עקומת הילברט
🔬💻🌊🎨📊✨
"""from spectra_codec import SpectraCodec
encoder = SpectraCodec()
original_filename = 'your_file.mzML'
output_filename = 'output.mzML'
encoder.encode_message_to_file(message, original_filename, output_filename)message = encoder.decode_message_from_file(output_filename)#, parser='pymzml', method='hilbert')
print("Decoded message:", message)Two public datasets carry SpectraCodec messages in every LCMS run:
- lbnl-metabolomics/20210915_JGI-AK_MK_506588_SoilWaterRep_final_QE-HF_C18_USDAY63680 — each run's full metadata (and the manuscript it belongs to)
- lbnl-metabolomics/ExoW3-NLDM — each run embeds the experiment's manifest, methods, protocols, compound identifications, and its strain's proteome fasta; the original documents (Excel, PDF, Markdown, fasta) are recoverable byte-for-byte from the spectra
Decode one:
python examples/decode_public_dataset/check_hf_message.pyHashes alone can't stop tampering — anyone can edit spectra and recompute
every checksum. SpectraCodec therefore embeds two Ed25519 signatures in a
provenance block of each signed message: one over the message payload (the
metadata and embedded documents), one over a digest of all acquired spectra
(everything except the carrier spectrum that holds the message). An attacker
who alters either the data or the metadata cannot re-create the signatures
without the signer's private key.
Verify a run against a published verification key:
python spectra_codec.py verify run.mzML --key spectracodec_verification_key.pemspectracodec_verification_key.pem in this repository is the LBNL
SpectraCodec verification key. Its fingerprint (SHA-256 of the raw 32-byte
public key) is:
SHA256:lbrclU9WYbrXjl/AWl8VOWFUEm5EQgGybgdSKZzTwfA
A valid signature from this key means the run was signed by the holder of the
corresponding private key and has not been modified since. Verifying without
--key uses the key embedded in the file itself, which only proves internal
consistency — not origin. To sign your own runs, mint a keypair with
python spectra_codec.py keygen, keep the private key out of any repository,
and publish the verification key and its fingerprint somewhere you control.
No package install needed — spectra_codec.py is a single-file library.
For a ready-made environment:
mamba env create -f environment.yml # creates spectra_codec_envCore dependencies: numpy, pandas, lxml, pymzml, pyteomics, psims.
(openpyxl is needed to open recovered .xlsx files in the examples.)
examples/decode_public_dataset/— download a file from the HF dataset and decode its embedded message; plus a decode-speed benchmark.examples/embed_experiment_files/— embed an experiment's supporting documents (manifest, methods, protocols, sequences) into every run of an experiment, with a unique ID per run and sha256-verified recovery of the original files. Start with its README.examples/manuscript/— notebooks and figure code from the SpectraCodec manuscript.