Exploring Spatio-Temporal Dynamics of Gene Expression using Transmodal Registration

Manan Lalit

(Jug & Tomancak Labs)

Dynamic changes in gene expression occur during early development

Embryo at 16 hpf*
Embryo at 20 hpf

?

* hours post fertilization

Building lineage trees with gene expression data requires bringing two datasets together

Live Embryo imaged with Light Sheet Microscopy: Nuclei, Membrane
In-situ Specimen at 16 hpf imaged with Confocal Microscopy: Nuclei, Pax3/7

+

=

Image Registration holds the answer

  • Intensity or Feature based
  • Global Transform based on Linear Models,
  • Local Transform based on Non-Linear, Elastic Models

Add Image

Feature - based
Intensity - based
(Source: MathWorks)

In-situ specimens for 50 genes at 3 developmental stages were imaged

Mette Handberg-Thorsager

Live imaging  & tracking for three embryos was performed

Mette Handberg-Thorsager

The number of cell nuclei is used as a measure for staging the in-situ specimen

Number of cell nuclei in in-situ specimens at 16 and 20 hpf

Selection of the correct time point is based on               maximum bi-partite matching

Cell nuclei in in-situ specimens are extracted

Feature Detection with Automatic Scale Selection, Lindeberg, T., International Journal of Computer Vision 
Live Embryo: Nuclei. Cell nuclei positions were fitted to an ellipsoid
In-situ Specimen: Nuclei. Cell nuclei positions were fitted to an ellipsoid

Fixation introduces shrinking of the embryo

Choosing a few corresponding nuclei helps estimate the underlying transform

Live Embryo: Nuclei, Membrane 
In-situ Specimen: Nuclei 

T=

Maximum Bi-partite Matching helps fine-tune the transform

Live Embryo: Nuclei
In-situ Specimen AFTER transformation: Nuclei 
In-situ Specimen BEFORE transformation: Nuclei
Live Embryo: Nuclei

T=

Lineage trees with gene expression data would allow cellular resolution, co-expression studies

In-situ Specimen imaged with Confocal Microscopy: Nuclei, pax3/7
In-situ Specimen imaged with Confocal Microscopy: Nuclei, pax6

Co-expression (Max z-projection ), obtained using ProSPr *

* Whole-organism cellular gene expression atlas, Vergara et al,  

Future: By-passing the hand-picking of corresponding nuclei

  • ">50" genes times  "3" time-points times ">15" replicates = >2250 in-situ images
  • Similar nuclei would have similar neighborhoods 
  • Explore elastic, non-rigid registration
Vladimir Ulman
16 hpf
20 hpf

Future: Adding Single Cell RNA-Sequencing data to the mix

Developmental Cell Lineage

Mette Handberg-Thorsager
Mette Handberg-Thorsager
Bruno Vellutini

Thank you for listening! Any questions?

Tomancak Lab

Marina Cuenca
Johannes Girstmair

Mette Handberg-Thorsager

Pavel Mejstrik
Stefan Münster
Giulia Serafini
Gabriella Turek
Vladimir Ulman
Bruno Cossermelli Vellutini

 

Jug Lab
Tim-Oliver Buccholz
Alexander Krull
Mangal Prakash
Deborah Schmidt

 

 

Jug/ Zechner Lab

Tobias Pietzsch

Jug/ Honigmann Lab

Anna Goncharova

Jug/ Zerial Lab

Nuno Pimpao Martins

Myers Lab

Matthias Arzt
Light Microscopy Facility

 

 

Computer Department