msms-compression

MS/MS Data Compression Package

Description

This Python package is designed for efficient compression of Mass Spectrometry (MS/MS) data. It is based on the MassComp algorithm, which is described in the following paper: https://doi.org/10.1186/s12859-019-2962-7

Version

0.2.0

Features

• Delta and Hex Encoding: Efficiently encodes m/z values and intensities to optimize the compression.
• Brotli Compression: Utilizes Brotli, a high-performance compression algorithm, offering superior compression ratios and speeds compared to gzip.

Installation

To install the MS/MS Data Compression package, run:

pip install msms-compression

Usage

The package includes the following main compressor classes:

• SpectrumCompressorUrl: Utilizes URL-safe Base64 encoding.
• SpectrumCompressor: Uses Base85 encoding.

• Note: The m/z values must be sorted in ascending order before compression, and contain only positive values.

Example:

from msms_compression import SpectrumCompressorF32

# Sample data
mz_values, intensity_values = [100.0, 101.0, 102.0], [10.0, 20.0, 30.0]

# Initialize the compressor
compressor = SpectrumCompressorF32()

# Compress data
compressed_data = compressor.compress(mz_values, intensity_values)
print("Compressed Data:", compressed_data)

# Decompress data
decompressed_mz, decompressed_intensity = compressor.decompress(compressed_data)
assert decompressed_mz == mz_values
assert decompressed_intensity == intensity_values

Compression Strategy Comparison

strategy	Compression Ratio	Compression Ratio Rank	URL Compression Ratio	URL Compression Ratio Rank	Compression Time	Compression Time Rank	Decompression Time	Decompression Time Rank
SpectrumCompressorLossy	5.952	1	5.023	2	0.030	5	0.008	4
SpectrumCompressorUrlLossy	5.579	2	6.926	1	0.030	4	0.007	1
SpectrumCompressor	3.890	3	3.299	6	0.053	7	0.010	6
SpectrumCompressorUrl	3.646	4	4.528	3	0.051	6	0.008	3
SpectrumCompressorGzip	3.148	5	2.658	7	0.023	2	0.009	5
SpectrumCompressorUrlGzip	2.951	6	3.665	4	0.022	1	0.007	2
SpectrumCompressorUrlLzstring	2.800	7	3.418	5	0.026	3	0.097	7

scan	strategy	original_size	compressed_size	url_encoded_size	compression_ratio	url_compression_ratio	compressed_time	decompressed_time
0	SpectrumCompressor	56124	14428	21139	3.88993623509842	3.299068073229576	0.053049564361572266	0.009985208511352539
0	SpectrumCompressorUrl	56124	15392	15401	3.646309771309771	4.528212453736771	0.051015615463256836	0.00789642333984375
0	SpectrumCompressorGzip	56124	17829	26236	3.1479050984351336	2.6581414849824667	0.02299976348876953	0.009003639221191406
0	SpectrumCompressorUrlGzip	56124	19020	19029	2.950788643533123	3.664879920121919	0.021996021270751953	0.007005453109741211
0	SpectrumCompressorUrlLzstring	56124	20041	20402	2.8004590589291953	3.418243309479463	0.026098012924194336	0.09739089012145996
0	SpectrumCompressorLossy	56124	9429	13884	5.952274896595609	5.022976087582829	0.030114173889160156	0.007976055145263672
0	SpectrumCompressorUrlLossy	56124	10060	10069	5.578926441351888	6.9261098420895815	0.030014991760253906	0.006910562515258789

The method compresses intensity values into two-character hexadecimal strings, offering 256 unique representations. This is a lossy approach, effectively reducing data size. Meanwhile, m/z values are compressed losslessly using delta encoding, maintaining their exact accuracy.