presentation.md

icon and intro slide

What can 50M cells teach us about gene networks?

Explain Gene networks

Explain scRNAseq

issues

is lack of ground truth and high dimension

Current methods

GRN inference methods, Large Cell Models claiming to map a model of the cell and doing so, can infer meaningful gene networks

LCM == ->

GENIE3 but also GNN based, other modalities, a bunch of models

train on one dataset only, need ground truth, slow to run and small networks, mostly on ODE generated fake data

other tools

transformer models

scPRINT

in depth overview

training

encoder and input

denoising & decoder

cell embedding and bottleneck learning

label prediction

abilities

showcase slide

Assessment

General overview

omnipath

MC. Calla

GWPS

issues in other models and zero shot abilities

denoising

batch effect correction and embedding

cell type prediction

analysing BPH

hubs and centrality

overlap and gene-gene comparison

finding communities

availability of scPRINT

usability example 1

usability example 2