Toxo Meeting Minutes
===
:::info
- **Date:** April 22, 2022, 1:00pm ET
- **Agenda**
1. Catch-up from April 8 meeting
2. Other updates
3. Goals for next meeting
- **Participants:**
- Shannon
- Gianni
- Alireza
- **Host:** Shannon
:::
:books: Catch-up from April 8
---
- Integration with napari
- Trying more preprocessing methods
- Try CellPose 2.0 model
- Napari tracking plugins
- Start writing?
- Overleaf https://www.overleaf.com/4332759976qqbckvmbbmnn
Virtual poster
- So definitely want something in a notebook / mybinder / napari format!
In particular:
- Silvia's paper (connection between spatial movement + fluorscence oscillation)
- DSAA 2019 paper
Any movement here?
:bust_in_silhouette: Other updates
---
Alireza
- No updates on Toxo
- Some cluster work
- A few interviews
Gianni
- CellPose 2.0 is a little unclear
- Start writing, learn more latex
- Tutorials on napari
:dart: Goals for next meeting
---
- Napari running
- Preprocessing
- Tracking + classification
- CellPose 2.0 and compare to PlantSeg
## Notes
Inserting figures in Overleaf https://www.overleaf.com/learn/latex/Inserting_Images
---
:::info
- **Date:** April 8, 2022, 1:00pm ET
- **Agenda**
1. Catch-up from Feb 11 (!) meeting
2. SciPy abstract
3. CellPose 2.0
4. Status of other papers
5. Goals for next meeting
- **Participants:**
- Shannon
- Gianni
- Alireza
- **Host:** Shannon
:::
:books: Feb 11 meeting catch-up
---
- Silvia + microscope resolution information [(Alireza answered)](/lTYTz1GwQkqI8NE2CiyoQw)
- Pretty much did a parameter scan over all the options in PlantSeg
- Getting "clairvoyant" segmentations: PlantSeg will identify a cell that doesn't appear, but then not identify it when it DOES appear deeper in the z-axis
- Rethinking some of the preprocessing
- Just a Gaussian filter seems to work reasonably well
- Still getting under-segmentation (some predicted masks that incorporate multiple actual cells)

- Mojtaba hard-coded some aspect, tailored specifically to the 3D dataset
- A lot of the de-noising took place during the reconstruction step
- Make napari a front-end for PlantSeg
:snake: SciPy abstract
---
**Accepted!** :clap:
Next steps?
Virtual poster
- So definitely want something in a notebook / mybinder / napari format!
:microscope: CellPose 2.0
---
- :newspaper: [Paper](https://www.biorxiv.org/content/10.1101/2022.04.01.486764v1)
- :bird: [Twitter](https://twitter.com/marius10p/status/1511415409047650307?s=20&t=umTVIG1CFKIWHYMrQqFKyQ)
- :spider_web: [Website](https://www.cellpose.org/)
Supposedly, version 2.0 not only makes training your own models easier (we do have supervised data!), but is also much better at anisotropic data. Might be worth a shot?
:writing_hand: Other papers
---
In particular:
- Silvia's paper (connection between spatial movement + fluorscence oscillation)
- DSAA 2019 paper
Any movement here?
:dart: Goals for next meeting
---
- Integration with napari
- Trying more preprocessing methods
- Try CellPose 2.0 model
- Napari tracking plugins
- Start writing?
- Overleaf https://www.overleaf.com/4332759976qqbckvmbbmnn
:closed_book: Tasks
--
==Importance== (1 *most* - 5 *least*) / Task / **Estimate** (# of hours)
## Notes
Funding flowchart for attending SciPy

---
###### tags: `Templates` `Meeting`
:::info
- **Date:** Feb 11, 2022, 1:00pm ET
- **Agenda**
1. Catch-up from Feb 4 meeting
2. SciPy abstract
3. Other updates
- **Participants:**
- Shannon
- Gianni
- Alireza
- **Host:** Shannon
:::
:books: Feb 4 meeting catch-up
---
- PlantSeg strategies
- CellPose
- Mojtaba + cleaned 3D data
- Silvia + microscope resolution information [(Alireza answered)](/lTYTz1GwQkqI8NE2CiyoQw)
- Sparse deconv paper
:snake: SciPy abstract
---
- https://docs.google.com/document/d/111LaBTtnMfwXMV7H01WFmUPf1VpWWDNQhoXSyzosZHI/edit?usp=sharing
- [The final document I submitted](https://docs.google.com/document/d/19PSaDyk0G9MuuxyNX9Z83L3LM-p4HvJAuw9TaKrhe4Y/edit?usp=sharing)
:dart: Other updates
---
- Napari plugin for toxo segmentation / tracking / phenotype classification
- Convert the SciPy paper to this
- Make a list of other major plugins available that kinda sorta do what ours does; we'll need to test some of these as part of our paper
## Goals for this week
- Pull out preprocessing steps from 2018 code https://github.com/quinngroup/toxoplasma-3DTracking
- Yeah, [this is definitely the preprocessing code](https://github.com/quinngroup/toxoplasma-3DTracking/blob/master/notebooks/3D_Tracking_detection_module.ipynb)
- How much preprocessing is built into PlantSeg?
~~- Edit and submit SciPy abstract~~
- Victoria has offered to edit!
:closed_book: Tasks
--
==Importance== (1 *most* - 5 *least*) / Task / **Estimate** (# of hours)
- [x] ==2== Ask Silvia about microscope resolution / magnification, and see if perhaps she knows pixel size
- [x] ==1== Start playing with PlantSeg, and in particular look into its preprocessing abilities
- [x] ==1== Edit SciPy abstract to include napari plugin; staged analytics that includes swappable preprocessing; performance improvements from the 2018 proof of concept
## Notes
- One of the bigger problems in the [2018 3D paper was locality](https://github.com/osbd/osbd-2018/raw/master/proceedings/S09209_6003.pdf)
- Centroids used from frame to frame
- Noisey centroids required filtering and additional processing
---
###### tags: `Templates` `Meeting`
:::info
- **Date:** Feb 4, 2022, 2:00pm ET
- **Agenda**
1. Catch-up from fall 2021
2. Plans for spring 2022
- **Participants:**
- Shannon
- Gianni
- Alireza
- **Host:** Shannon
:::
:books: Fall 2021 Review
---
- Segmentation strategies using PlantSeg
:dart: Spring 2022 Goals
---
- Graduate Gianni (May or August)!
- Check out CellPose
- Get a paper out
- [JOSS](https://joss.theoj.org/)
- Silvia's paper (connection between spatial movement + fluorscence oscillation)
- DSAA 2019 paper
- Computational methods for exploring spatial movement + fluorescence oscillation
- [:heavy_check_mark:] SciPy paper https://www.scipy2022.scipy.org/talk-poster-presentations . Abstract deadline: Feb 22
- Cleaning up images / noise in 3D images
- Look into sparse convolutions
## Goals for this week
- What kind of segmentation? is plantseg the way to go, or some other pre-packaged / pre-trained 3D segmentation?
- Figure out a preprocessing pipeline
- Start designing for JOSS submission
- [:heavy_check_mark:] Draft for SciPy abstract (due Feb 11) [Google doc](https://docs.google.com/document/d/111LaBTtnMfwXMV7H01WFmUPf1VpWWDNQhoXSyzosZHI/edit?usp=sharing)
:closed_book: Tasks
--
==Importance== (1 *most* - 5 *least*) / Task / **Estimate** (# of hours)
- [x] ==1== Find out if Mojtaba has an archive of the cleaned 3D data
- Seems unlikely
- [x] ==1== Start working on a SciPy abstract(s) to workshop next week [Google doc](https://docs.google.com/document/d/111LaBTtnMfwXMV7H01WFmUPf1VpWWDNQhoXSyzosZHI/edit?usp=sharing)
- [ ] ==2== Ask Silvia about microscope resolution / magnification, and see if perhaps she knows pixel size
- [ ] ==2== Read sparse deconv paper https://www.nature.com/articles/s41587-021-01092-2
## Notes
<!-- Other important details discussed during the meeting can be entered here. -->
Meetings will normally be every other week on Fridays at 1pm
Idea of signal propagation across space and time
- within a single time point (z-slices), find sources of fluorescence that persist in multiple consecutive z-slices
- across multiple time points, find objects that persist across multiple consecutive time points