Johannes Köster
2016
Fluorescence in-situ hybridization:
Problem:
needs 1 color per transcript
...
Problem with raw counts:
20% loss and misidentification
Known error probabilities:
1→0 error: 10%
0→1 error: 4%
Goal:
Pr(0→1) Pr(1→0)
simulated
hybridization
Published dataset:
Question:
Are expression profiles influenced by batch effects?
cell size
cell position
batch
Term | expected | observed | size |
---|---|---|---|
response to temperature stimulus | 0.36 | 3 | 6 |
cellular response to heat | 0.36 | 3 | 6 |
negative regulation of endopeptidase activity | 0.12 | 2 | 2 |
second-messenger-mediated signaling | 0.12 | 2 | 2 |
positive regulation of protein kinase B signaling | 0.12 | 2 | 2 |
regulation of peptidase activity | 0.12 | 2 | 2 |
pos. regulation of reactive oxygen species metabolic process | 0.12 | 2 | 2 |
controlled FDR at 5% with Benjamini-Yekuteli
Bayesian model for gene expression analysis on MERFISH data:
Simulation:
can correct for biases in MERFISH data
Application:
applied to characterize batch effects
Shirley Liu
Myles Brown
Bo Li
Peng Jiang
Eric Severson