diff --git a/README.md b/README.md index 466d962..44de5d3 100644 --- a/README.md +++ b/README.md @@ -1,6 +1,6 @@ # Kun-peng Kun-peng Logo -[![](https://img.shields.io/badge/doi-waiting-yellow.svg)]() [![](https://img.shields.io/badge/release%20version-0.6.15-green.svg)](https://github.com/eric9n/Kun-peng/releases) +[![](https://img.shields.io/badge/doi-waiting-yellow.svg)]() [![](https://img.shields.io/badge/release%20version-0.7.0-green.svg)](https://github.com/eric9n/Kun-peng/releases) Here, we introduce Kun-peng, an ultra-memory-efficient metagenomic classification tool (Fig. 1). Inspired by Kraken2's k-mer-based approach, Kun-peng employs algorithms for minimizer generation, hash table querying, and classification. The cornerstone of Kun-peng's memory efficiency lies in its unique ordered block design for reference database. This strategy dramatically reduces memory usage without compromising speed, enabling Kun-peng to be executed on both personal computers and HPCP for most databases. Moreover, Kun-peng incorporates an advanced sliding window algorithm for sequence classifications to reduce the false-positive rates. Finally, Kun-peng supports parallel processing algorithms to further bolster its speed. Kun-peng offers two classification modes: Memory-Efficient Mode (Kun-peng-M) and Full-Speed Mode (Kun-peng-F). Remarkably, Kun-peng-M achieves a comparable processing time to Kraken2 while using less than 10% of its memory. Kun-peng-F loads all the database blocks simultaneously, matching Kraken2's memory usage while surpassing its speed. Notably, Kun-peng is compatible with the reference database built by Kraken2 and the associated abundance estimate tool Bracken1, making the transition from Kraken2 effortless. The name "Kun-peng" was derived from Chinese mythology and refers to a creature transforming between a giant fish (Kun) and a giant bird (Peng), reflecting the software's flexibility in navigating complex metagenomic data landscapes.