A deep learning-based sequence classifier applied to 16S rRNA.

This repo builds further on the findings of my bachelor's project with the goal of creating a highly accurate, easy-to-use and easy-to-interpret sequence classification tool for the 16S rRNA gene but which can be applied and quickly re-trained to any marker gene.

Things to work on:

• additional data augmentation strategies (sequence cropping)
• creating combined model that takes in both k-mer and one-hot encoded sequences
• reduce training memory requirements by building a TensorFlow input pipeline (tf.data)
• create a BERT-like multi-class & multi-label classifier
• do something about sequence entries with partially missing labels and how to weight the model around this problem...