From confocal images into computer analyzable, cellular resolution spatial expression data

1
1
3
From confocal images into computer analyzable,
cellular resolution spatial expression data
The xyz-positions of the blastoderm nuclei and
the expression intensity of two genes around each
nucleus are computationally extracted from a
confocal image (A) and converted into a computer
readable data table (B). This text-file of
compressed expression data is the basis for all
the subsequent analyses and visualization tools.
Sample preparation gtgt Fluorescent in situ
stains Imaging gtgt Confocal image
stacks Nuclear segmentation gtgt PointCloud
data converts 300 Mb image into 1 Mb text
file Registration gtgt Virtual PointCloud
data (see below) Visualization
tools gtgt PointCloudXplore
id,x,y,z,Nx,Ny,Nz,Da,Db,Vn,Vc,eve,ftz,hb,kni,kr,rh
o,slp1,sna,tll,croc,fkh,twi,trn,gt
1009,142.45,124.288,38.4751,-0.22
86,0.37424,-0.89871,0,0,317.559,864.635,13.7058,81
.2576,14.1843,21.9375,4.9696,7.9223,10.912,6.5325,
18.3249,17.3787,18.499,10.9727,38.7896,80. 2018,15
9.302,123.878,164.709,-0.10815,0.3775,0.91967,0,0,
308.755,865.85,70.6029,10.0973,25.2076,17.8543,8.8
861,33.3451,13.8321,58.1229,17.7753,27.252,20.3735
,70.5197,10.4731,2 3027,111.273,115.159,158.502,-0
.23647,0.3794,0.8945,0,0,268.984,685.818,12.2596,5
.7213,8.0693,69.3157,6.614,28.261,70.3689,58.5122,
15.6545,22.3175,15.5599,63.354,45.7422,40.88 4036,
341.931,30.2893,75.7972,0.18502,-0.94545,-0.26812,
0,0,231.338,537.361,19.9463,53.7362,65.8337,21.370
2,8.2404,19.4855,8.0923,5.6487,25.3556,18.5914,22.
0013,10.0438,22.6887 5045,168.5,25.747,67.1179,-0.
10967,-0.90025,-0.42134,0,0,224.052,630.261,51.517
,8.0767,22.2996,23.0654,11.2735,46.5546,12.3596,5.
6873,15.8975,24.598,25.5458,9.9865,16.8356,20. 605
4,119.682,108.526,168.746,-0.19491,0.21631,0.95667
,0,0,170.923,505.484,67.6004,6.7196,10.9492,24.875
7,7.5595,58.568,66.6675,36.9671,23.4779,15.9494,18
.4064,59.0062,17.4559, 1,175.229,155.285,55.1238,-
0.14639,0.80211,-0.57896,0,0,285.682,582.293,10.24
66,89.2554,23.3279,18.3,10.1307,16.224,13.789,5.07
49,15.5772,21.7618,22,15.5339,34.8946,16.5044 ...
BDTNP Bioimaging Database
2
4
Drosophila embryo as a model for developmental
basis of evolutionary change
PointCloud explore facilitates the exploration of
multidimensional expression datasets from
multiple angles
Cellular resolution alignment of multiple embryos
allows comparison of multiple genes in the same
morphological framework
Drosophila melanogaster embryo is a common model
in developmental biology. Its basic bodyplan is
determined during blastoderm stage by a cascade
of regulatory interactions that read the maternal
inputs into spatial information.
embryo view
3D scatterplot
The individual PointClouds each contain the
information of only two gene products per cell
for one embryo. Moreover, the equivalent cells in
different embryos are in slightly different
positions. To compare the spatial and temporal of
many genes, we find the equivalent cells in
multiple PointClouds by using the spatial
information of one gene as a registration marker
to align them into a single Virtual PointCloud.
Examples of various views available in
PointCloudXplore
This converts genetically identical cells into
different cell types. Hence, to understand genome
function, we need to record the development of
local expression differences in a whole
developing organism at a cellular resolution.
PointClouds
cell magnifier
parallel co-ordinate view
unrolled view
Problem Approach
A screenshot of 3D surfaces over an unrolled
view and two control interfaces
Analyze 3D expression patterns of multiple genes
expressed in a blastoderm embryo to find
interactions between the genes
14000 genes translated in a selforganizing
manner into a stereotypical organism
Virtual PointCloud
Both single PointClouds and Virtual PointClouds
can be visualized with PointCloudXplore, which
can display the multidimensionsal expression data
both on spatial and abstract dimensions. The
embryo view displays the PointCloud data as
visual representation of the embryo. The whole
blastoderm can be viewed simultaneously as an
unrolled view. The relative expression levels of
all the genes in different cells can be viewed as
parallel co-ordinates or in a 3D scatter plot for
a group of cells, or in an individual cell
magnifier. To better distinguish between
expression levels of overlapping genes, the
levels can be displayed as novel 3D surfaces over
a 2D view. The various control interfaces allow
the user to choose the views and the genes and
their display intensities to display from the
total data set.
input
edge
node
network
output
6
5
Computational analysis shows nuclear density
changes in stage 5 Drosophila melanogaster
blastoderm through time
3D analysis of gene expression reveal
complexities not seen in 1D analyses
ftz mRNA
eve mRNA
Pattern elements of anterior-posterior regulators
of bodyplan eve, ftz, gt, hb and kr show
differential development on dorsal and ventral
sides of the embryo. This movement is different
for different elements and genes. The genes also
have different temporal profiles for
anterior-posterior pattern changes (not
shown). These differences are partially due to
dorso-ventral differences in blastoderm
morphogenesis. The dorso-ventral density
differences (see left) are caused by nuclear
movement towards denser patches. Because embryos
are three-dimensional structures, this causes
also anterior-posterior movement of nuclei. Since
anterior-posterior movement is greater dorsally
than ventrally, part of the higher anterior
movement of dorsal than ventral posterior
elements is caused by nuclear movement
However, the difference between predicted late
positions (black stippled line) and real late
positions (red line) indicates that some of the
dorso-ventral differences in pattern formation
are caused by direct inductive interactions
between genes, rather than as a by-product of
morphogenesis. This proves that to correctly
model the pattern formation in Drosophila
embryos, we need cellular resolution 3D data and
temporal correspondences between individual
nuclei.
If we want to accurately map the temporal
expression profiles of each individual cell, we
need to know if cells change position over time.
We have divided 896 PointClouds into different
age cohorts to measure any expression shifts,
when we discovered that the local nuclear
densities increase dorsally (red) and decrease
anteriorly (blue). This suggests that the nuclei
must move. Our live cell time series data shows
that this is the case.
gt mRNA
hb mRNA
kr mRNA
early border position late border position
late border position projected from cell
movement alone