Image analysis tool used for tracking localized growth of hypocotyls using high-throughput cloud computing. This program uses a variety of machine learning algorithms to identify a hypocotyl, extract the growth region, form a midline, then quantify the elemental growth rate in which cellular elongation occurs at many locations along that midline.
Used BFG to remove all objects > 500K in size [07-09-2019]
A) What is this program for?
Learning algorithms: predict contour, predict ending of midline
Image analysis: select points along seedling
Quantification: Localized measurements of midline, model growth of growing seedlings
Miscellaneous features: late-decision segmentation for error checking, ease of loading data
Output Data: models full developmental process of a growing, de-etiolating hypocotyl
B) How does this program work?
User stores time-lapse image stacks of a single type in individual folders
A full experiment is a set of individual image stacks in a parent folder
Multiple experiments can be loaded as well
C) Instructions for using HypoQuantyl can be found in HOWTO.md
- Sample data can be found in Sample Data
- Lowest usable MATLAB version
- Run scripts for toolbox dependencies (like from QuantDRaCALA)
- Maybe something about CyVerse if necessary?
- (to-do) Install program as Matlab executable
- (to-do) Program runs on CyVerse
02/05/2018 - HypoQuantyl Version 0.1 [ date created ] 02/06/2018 - HypoQuantyl Version 0.1 [ name chosen ] 02/22/2018 - HypoQuantyl Version 0.3 [ class structure finalized ] 05/24/2018 - HypoQuantyl Version 0.4 [ midpoint-normalization method ] 08/02/2018 - HypoQuantyl Version 0.5 [ routes method --> curves method ] 12/05/2018 - HypoQuantyl Version 0.6 [ optimized image read/write ] 02/18/2019 - HypoQuantyl Version 0.7 [ learning-based segmentation framework ]
Julian Bustamante, Cellular and Molecular Biology Program (jbustamante@wisc.edu)
University of Wisconsin - Madison
Department of Botany
Nathan Miller, Senior Scientist (ndmill@gmail.com)
University of Wisconsin - Madison
Department of Botany
MIT license found in LICENSE
- Guosheng Wu
- Nathan Miller
- Segmentation with Convolution Neural Net (CNN)
- CNN Architecture
- Feature analysis
- CNN Architecture
- Junction Finder with CNN
- Click Method
- Slice Method
- Click Method
- Midline Region Splitter and Feature Identifier
- Track patches between frames
- Track patches between frames
- Kinematics Analysis
- Elemental Growth Rate methods
- Elemental Growth Rate methods
- Validation Methods
- Manual methodology
- Manual methodology
- Deployment to CyVerse
- Compile to use with MCR
- Synchronization with Open Science Grid
- Compile to use with MCR
- True Application Testing
- CRY1 mutants
- Cell Wall mutants
- CRY1 mutants
Class | Description |
---|---|
HypoQuantyl | Initializes program and loads Experiment folders |
Experiment | Loads folders containing multiple image stacks |
Genotype | Loads image stacks containing growing seedlings |
Seedling | Represents a single seedling throughout a time-lapse |
Hypocotyl | Represents the hypocotyl portion of a single seedlings |
CircuitJB | Main contour of a hypocotyl, segmented by anchor points |
Route | Individual segments around a hypocotyl's contour |
Algorithm | Description |
---|---|
FindHypocotyl | Identifies region on a seedling containing the hypocotyl |
TrackHypocotyl | Keeps track of hypocotyl during de-etiolation process |
FindGoodFrames | Runs error-checking of each frame for each Seedling |
IntegrationAlgorithm | Integration method for measuring growing Hypocotyl |
Infrastructure | Description |
---|---|
Create accounts | University Network, CyVerse, HTCondor |
Set up submit files | Holds MATLAB MCR, HypoQuantyl, Data, I-Commands |
Run Test Data | Run HypoQuantyl on Sample data on cloud computing environment |