Contact

80-Fold Software Performance Improvement

Azati significantly accelerated the client’s DNA sequence processing software by identifying and optimizing a critical bottleneck in the FASTAptamer toolkit. The team rewrote the core clusterization logic from Perl to C++, achieving an 80x reduction in total execution time and a 1,000x improvement for the embedded Levenshtein algorithm.

Discuss an idea

All Technologies Used

C++
C++
Perl
Perl

Motivation

The client approached Azati with a request to dramatically enhance the performance of their data processing pipeline, which handled vast amounts of sequencing data. The key objective was to reduce the time required to process biological sequences without compromising output accuracy.

Main Challenges

Challenge 1
FASTAptamer Performance Limits Speed

The client’s software took approximately 48 hours to process sequencing data due to a bottleneck in the FASTAptamer toolkit, severely impacting research timelines. Azati proposed to analyze the entire pipeline, identify the most resource-heavy step, and optimize it with low-level performance engineering.

Challenge 2
Inefficient Sequence Clustering in Perl

The specific clusterization step using the Levenshtein algorithm, implemented in Perl, was highly inefficient and caused long delays in the workflow. Azati suggested rewriting this logic in C++ to leverage faster execution and memory handling capabilities, drastically improving performance.

Key Features

  • Accelerated clusterization logic: Replaced slow Perl-based Levenshtein logic with high-speed C++ implementation.
  • Massive performance gain: Achieved a 1,000x improvement in algorithm execution and 80x overall pipeline speedup.
  • Open-source contribution: Enhancement was merged into the official FASTAptamer package, benefiting the broader bioinformatics community.

Our Approach

Pipeline Analysis
Analyzed the client's DNA sequencing pipeline to pinpoint performance bottlenecks.
Bottleneck Identification
Identified that the FASTAptamer's clusterization program, particularly its use of the Levenshtein algorithm, was consuming the majority of processing time.
Benchmarking and Root Cause Analysis
Benchmarked the Perl implementation and determined that its inefficiency stemmed from language limitations in high-throughput operations.
Algorithm Rewriting in C++
Rewrote the Levenshtein algorithm in C++ to improve performance, exploiting native memory management and compilation advantages.
Integration and Validation
Integrated the optimized C++ version into the client's software and conducted tests to ensure result consistency and correctness.
Contribution to Open Source
Submitted the improved algorithm to the official FASTAptamer repository, which was accepted and included in version 1.0.12.

Project Impact

Faster Data Processing: Reduced total runtime from 48 hours to just 30.5 minutes, drastically improving research productivity.

Validated Results: Output results remained consistent, ensuring scientific integrity and reproducibility.

Community Benefit: The optimization was accepted into the mainstream toolset, making it available to all FASTAptamer users globally.

Ready To Get Started

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.