Deep Learning-Based Morphological Profiling for Rare Disease Genomic Medicine

Explore deep learning-based morphological profiling for rare disease genomic medicine in this comprehensive lecture by Wolfgang Pernice from Columbia University Irving Medical Center. Delve into the challenges of functionally interpreting genetic variations in disease contexts and learn how patient cell profiling offers powerful solutions. Discover the potential of deep representation learning in unlocking cellular morphology as a cost-efficient and rich domain of cell biology. Examine approaches to overcome out-of-distribution generalization challenges, including a novel method based on generative interventions. Gain insights into batch-effect correction techniques and their applications in rare disease genomic medicine. Cover topics such as variant filtering, imaging data sets, confusion matrices, feature embeddings, causal inference, and UMAP categorization. Understand the implications of this research for addressing roadblocks in rare disease genomics and beyond.

Introduction
Context
Variant filtering
Application
Imaging
Data set
Confusion Matrix
Feature embeddings
Causal inference
Generalisation
Batch License Score
Reverse gradient reversal layers
Umap categorization
Uncorrelated data
Results
Thank you
Inverse gradient lights
Gan components
Class probabilities