The Data Blog

In November, I released the version 2.64 of SPMF with 5 new algorithms and 2 new tools, among other improvements, but I still had several other pattern mining algorithms waiting to be released after more tests.

Thus, today, I make a new release of SPMF for December 2025, called version 2.64b. This release has several new algorithms:

• The HMP-SA algorithm for discovering compressing itemsets in a transaction database using a simulated annealing approach (Chen, E. et al., 2026).
• The HMP-HC algorithm for discovering compressing itemsets in a transaction database using a hill-climbing approach (Chen, E. et al., 2026).
• the GENMAX algorithm for mining frequent maximal itemsets from a transaction database (Gouda et al., 2005)
• the DIC algorithm for mining frequent itemsets in a transaction database using dynamic itemset counting (Brin et al. 1997)
• the Talky-G and Talky-G-Diffset algorithms for mining frequent generator itemsets in a transaction database (Szathmary, 2009)
• the iMEFIM algorithm for high utility itemset mining, a variation of EFIM that can be faster but can consume more memory (Nguyen et al., 2019)
• the PUCPMiner algorithm for high utility itemset mining, a variation of FHM with a potentially tighter upper bound (Patel et al., 2022

Moreover, there is a new Pattern Diff Analyzer tool that allows to compare two pattern files to find contrast patterns. This tool works letting the user first select two files containing patterns:

And then we can compute the differences between these files using different contrast methods. For example, we can identify all patterns that are in the first file but not in the second, among multiple other options for identifying contrast patterns:

That is all I wanted to share about this new version! Hope that you will enjoy this new release of the SPMF pattern mining software. So far we have released 18 new algorithms this year! And more features are coming soon…

Today, I am pleased to share that our team has published a new algorithm called GMP for the compression and analysis of protein sequences. The paper will appear next month at BIBM 2025. Here is the full reference:

Nawaz, M. Z., Nawaz, S., Fournier-Viger, P, Niu, X., Li, M. (2025). A Multipurpose Protein Compressor based on MDL and Genetic Algorithm. Proceedings of BIBM 2025.

And you can watch the video of the presentation on Youtube:
Presentation (18 minutes)

Abstract

The rapid expansion of protein sequence databases has created challenges for efficient storage, transmission, and analysis. Unlike genomic sequences with only four nucleotide bases, proteins are composed of twenty amino acids, making compression more complex. Existing specialized protein compressors, such as AC, AC2, and CPM-FCM, have achieved promising performance but still face limitations, including high computational cost, low adaptability, and limited biological interpretability. This paper introduces GMP (Genetic algorithm-based MDL Protein compressor), a novel protein compression framework that leverages the Minimum Description Length (MDL) principle with a genetic algorithm to discover optimal patterns of amino acid subsequences (kAA-mers). Experimental results demonstrate that GMP attains compression performance comparable to state-of-the-art methods while additionally supporting tasks such as classification and clustering—capabilities absent from traditional protein compressors. This makes GMP not only an efficient compression framework but also a biologically interpretable tool for protein sequence analysis. GMP is available at github.com/MuhammadzohaibNawaz/GMP.

Index Terms—Protein sequences, Compression, Genetic Algorithm, Minimum Description Length, kAA-mers

In summary

GMP was designed not only to compress protein sequences but also to provide insights into their structure through the discovery of meaningful subsequence patterns. By integrating MDL with a genetic algorithm, it strikes an effective balance between compression quality and interpretability. One of the unique strengths of GMP is that it can simultaneously serve multiple purposes: compression, classification, clustering, and pattern discovery—functions rarely combined in a single framework. Here is the main flowchart from the paper:

We will release the paper soon after it is published next month at BIBM 2025.

Today, I wrote some code to visualize the relationships between the algorithms offered in the SPMF open-source pattern mining library. Here is the graph of all algorithms in SPMF (excluding tools) that have the same input and output type:

We can see a few big clusters such as high utility itemset mining algorithms:

Frequent itemset mining algorithms:

Closed itemset mining algorithms:

Episode rule mining algorithms:

Frequent sequential pattern mining algorithms:

Sequential rule mining algorithms:

Frequent episode mining algorithms:

If we only display the algorithms using edges if they have the same input file type (but may not have the output), the graph is different:

Now, there is a huge cluster for itemset mining and association rule mining:

And there is a big cluster for sequence mining algorithms:

And a smaller cluster for episode mining:

and a big cluster for high utility pattern mining:

The tool that I use to draw these pictures is the AlgorithmGraphVisualizer, a new GUI tool, which will be offered in the next version of SPMF. It allows to visualize algorithm relationships with different options and export to PNG. Here is the current GUI interface:

Hope that this has been interesting! The new version of SPMF should come out probably next week!