Receive and analyse new data set from Emmons Lab

[slarson]

Take a look at following issues:

• [x] Straightening process relative to existing data set
• [ ] Scaling new data set relative to existing data set -- with goal of having 3D representation of both data sets on top of each other with nuclei from worm 1 in red and worm 2 in blue with boundary of worm as overlapping as possible
• [ ] Manual label with minimal set of cells identified on existing data set
• [ ] Select a dozen cells that obviously the same in both data sets (if possible)
• [ ] Refer to the Peng paper on atlas about ideal next steps to create reference space that allows the new data set to be compared against the old.
• [ ] Define the way we are going to represent points in the space (e.g. probabilistically or an average?)

[EST09]

A 3D digital atlas of C. elegans and its application to single-cell analyses

• used DAPI to stain all nuclei then GFP (myo-3:GFP transgene) to label the nuclei of the 81 body wall muscle cells and 1 depressor muscle cell. These nuclei were fiducial markers. (-also had target set for which they use mCherry) -straighten worm - detects the principal curve or 'backbone' that represents the anterior-posterior line . Generates one-pixel separated planes orthogonal to the backbone and restacks them along a straight anterior-posterior line, making the planes parallel to each other . Because each local transform is a rigid rotation, the resolution of the straightened image is merely a function of the subpixel interpolation error when a straight line is rotated in space. Thus, the overall resolution loss is naturally minimized. -segmentation -used VANO to annotate (I got this from the authors already) -With a set of K stacks they decided on which was the reference stack, named C, by which stack minimised the sum of the squared differences of the 357 or so nuclei centers between C and every other stack -mapped every stack to C via affine transform -The final atlas is the ensemble of the transformed nuclei positions for each named nucleus in the coordinate system of the reference stack (ensemble? - how Willard Gibbs in 1878, an ensemble (also statistical ensemble or thermodynamic ensemble) is an idealization consisting of a large number of mental copies (sometimes infinitely many) of a system, considered all at once, each of which represents a possible state that the real system might be in.

Statistical analysis:

• looked at SD of nuclei centres (AP, LR, DV directions)
• used this to find their invariant nuclei

To map a subject to the atalas:

• To annotate markers: bipartite matching algorithm to find the best match between the 82 marker nuclei of C and S under the given transform, that is, that minimizes the sum of the Euclidean distance between corresponding nucleus centers.
• Then: took the affine transformation, tGFP, that minimized the difference between S and C with respect to all 82 GFP-labeled nuclei and used the now-annotated marker nuclei in S to triangulate the remaining cell nuclei in the DAPI channel
• computed the normalized distance between a nonmarker nucleus to be annotated and its nearest marker nuclei (one anterior and one posterior nucleus in each of the four muscle bundles) and used this metric to find the best bipartite matching between nuclei in S and a subset of those in the atlas (why- ) -then examined the AP, DV and LR relationship in our initial labeling of S to see whether there was any assignment that seemed to substantially conflict with the invariants of the AP, DV and LR graphs derived from the atlas

[EST09]

Need to do:

1. How does straightening affect scaling? How do pixel sizes change?

2. Need to ask Scott:

• Did they record their parameters for Z stack creation?
  • At what distance apart did they set the Z stacks?
  • At what point did they decide to start the Z stacks and at what point did they stop?
• If this is not available - talk about possible staining of a certain set of cells?

3. Cluster neurons of worms

4. Then go through Long paper and think about space