wasineer-dev / kdpp Goto Github PK

k-DPP for subset selection using VLMC distances

This repository contains an example on how to use DPPy and Tensorflow to sample exactly from k-DPP and evaluate log probability. We use Nextflow to implement a pipeline to compute VLMC pairwise distances between sequencing reads. To use the Nextflow pipeline, you have to install the VLMC library dvstar.

Input: Pairwise distances between VLMC

Output: Samples from k-DPP

• Nextflow
• BBTools such as bbduk.sh and reformat.sh
• dvstar to generate pairwise distances between VLMC
• DPPy
• Tensorflow

Improve conversion from distance to similarity and implement Gibbs sampling.

Alexander Schliep, Joel Gustafsson for help with dvstar