Relevant new features from scikit-bio 0.7.4

Developer Discussion
1

Hello Q2 team,

This is Qiyun from the scikit-bio team. @gregcaporaso kindly pointed me to this discussion board. We recently released version 0.7.4. Some new features may be of your interest because they may improve existing features of QIIME 2. Specifically:

  1. We have significantly improved computational performance of several core algorithms: PERMANOVA, Mantel, PCoA and PERMDISP, using modern techniques such as array API, Numba kernel, and GPU (Nvidia or AMD). A documentation page has been added to explain how to maximize performance of these.
  2. We now have native Python re-implementations of ANCOM-BC and ANCOM-BC2 that are hundreds of times faster than the original R package and consume less than 5% memory, while producing identical results. A paper describing these implementations has been published in Bioinformatics.

Just for your information. We are committed to support Q2 and the ecosystem. Let me know any questions or thoughts!

6 Likes