-
Notifications
You must be signed in to change notification settings - Fork 7
4 Registration
This stage performs registration on the normalized input. ExSequence does affine and thin plate spine registration. The registration method is presented in Murray at al 2015 in Cell. It uses a combination of 3D Harris Corner detection as well as 3D SIFT. The algorithm calculates correspondences between the reference round and the other rounds saving the results in a file:
globalkeys_basename-downsample_round###.mat
This allows for quick access to the features for further processing. After verifying the correspondences, a Thin Plate Spine is calculated to allow for fine grained corrections in the warping. For more information on the registration step see registrations's dedicated wiki.
For each round a summedNorm directory is created containing a .mat file containing a list of keypoints which are used to warp the rounds to the reference round (chosen in the parameters file). For the reference round, symlinks are created for each color channel (appending _affine to the file name) to the files produced by the normalization step. For the other rounds new images are created which are registered to the reference round.
- 1 Setup Cluster
- 2 Color Correction
- 3 Normalization
- 4 Registration
- 5 Puncta Extraction
- 6 Base Calling
- Performance Profile
- Example Batching
- Summary Reporting (link to page) / Analysis (link to page) All of above would be bullet points but thought each folder with useful scripts in the repository would have its own page with example usages and descriptions for the files under the given directory. Unsure on this as haven't used anything under these directories directly myself (i.e not through runPipeline) apart from tests/perf-profile/summarize-stat-logs.sh. Should any of these have a separate page? Note have not made pages for this yet as not sure how many should be made or if this break down makes sense.
- Troubleshooting