Master's Thesis Research

MPhil in Computational Biology with the Liston Laboratory

0. Data availability

The flow cytometry dataset for this project was provided by J. Neumann, Dr. S. Humblet-Baron, and Dr. L. Yshii (KU Leuven). The participant population comprised 143 males and 158 females aged between 40-83 years old. Participants were further characterized by clinical measures including age, age at diagnosis, and Mini‐Mental State Examination (MMSE) score at the time of diagnosis.

Initial data processing was performed by Dr. O. Burton (University of Cambridge) using FlowJo™ Software (BD Life Sciences). Processing steps included the removal of artifacts, doublets, and dead cells and compensation correction by Autospill.

1. Data pre-processing

1.1 Statistical analyses (linear mixed modeling)

1.2 Creating ML-friendly files

1.2.1 Pre-processing

1.2.2 Test/train/validation split

1.2.3 Data transformations

1.2.4 Data augmentation

2. Machine learning classifiers

2.1 Cell level models

2.2 Sample-level models

3. UMAP+ML classifier pipeline

4. Cell-subset gating