All-Hands Meeting Minutes === October 28, 2024 --- :::info - **Time:** 12:30pm ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr - Alireza ::: Data on GACRC: `/project/spqlab/lab_shared/ornet/Confocal_3D_2024_ND2s/` (everything here should be current, as of 10/11) Amr's location: `/lustre2/work/spqlab/` ISBI Google doc: https://docs.google.com/document/d/10AFLGFnK6aaVcHIDjybOeCsXr-6eMONgLYW3u65wnlw/edit LaTeX paper: https://github.com/quinngroup/OrNet_ISBI_2025 Project timeline: https://github.com/orgs/quinngroup/projects/3/views/4 Initial analysis report: https://docs.google.com/document/d/1VI_CPuzqJxgOJ4jTQNzr6QFVsLJ-xA9fPdidHkmoSxg/edit ### Checklist to Completion **ISBI submission deadline extended to Oct 31** Checklist to completion: Analysis: - [ ] False positive / false negative / true positive / true negative consistency analysis - [ ] 2-3 top features corresponding to classification - [ ] PCA embedding - [ ] Compute a rough GFP signal strength value for each cell (could do a non-maximal z-stack suppression, then sum up the GFP pixels for a cell) - [ ] Use this as a post-hoc analysis tool when investigating the results of the consistency analysis - [ ] Add a flag to each cell to indicate whether manual inspection yielded any GFP signal or not - [ ] Use this to train a model eliminating all cells with no detectable levels of GFP - [ ] Label cells in a field of view uniformly - [x] Train a classifier that ignores cells with no detectable GFP signal - [ ] Train a classifier of wildtype vs uninfected Experimental: - [x] Final uninfected dataset - [x] Final infected dataset - [x] High-res images of each "class" for a figure - [x] Amr + Alireza work together on annotating remaining data Written: - [ ] Computational methods - [ ] Biomolecular / infectious disease justification for this study - [ ] Figures - [ ] Data example (each class) - [ ] 3D Embeddings - [ ] Table of classification results / top features - [ ] Renderings of "exemplar" versions with features in play - [ ] Results - [ ] Discussion and Conclusions ### Todo Shannon: - [ ] Register on conference submission site Everyone: Next meeting: October 23, 2024 --- :::info - **Time:** 2:00pm ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr - Alireza ::: Data on GACRC: `/project/spqlab/lab_shared/ornet/Confocal_3D_2024_ND2s/` (everything here should be current, as of 10/11) Amr's location: `/lustre2/work/spqlab/` ISBI Google doc: https://docs.google.com/document/d/10AFLGFnK6aaVcHIDjybOeCsXr-6eMONgLYW3u65wnlw/edit LaTeX paper: https://github.com/quinngroup/OrNet_ISBI_2025 Project timeline: https://github.com/orgs/quinngroup/projects/3/views/4 Initial analysis report: https://docs.google.com/document/d/1VI_CPuzqJxgOJ4jTQNzr6QFVsLJ-xA9fPdidHkmoSxg/edit ### Checklist to Completion **ISBI submission deadline extended to Oct 31** Checklist to completion: Analysis: - [ ] False positive / false negative / true positive / true negative consistency analysis - [ ] 2-3 top features corresponding to classification - [ ] PCA embedding - [ ] Compute a rough GFP signal strength value for each cell (could do a non-maximal z-stack suppression, then sum up the GFP pixels for a cell) - [ ] Use this as a post-hoc analysis tool when investigating the results of the consistency analysis - [ ] Add a flag to each cell to indicate whether manual inspection yielded any GFP signal or not - [ ] Use this to train a model eliminating all cells with no detectable levels of GFP - [ ] Label cells in a field of view uniformly - [ ] Train a classifier that ignores cells with no detectable GFP signal - [ ] Train a classifier of wildtype vs uninfected Experimental: - [ ] Final uninfected dataset - [ ] High-res images of each "class" for a figure - [ ] Amr + Alireza work together on annotating remaining data Written: - [ ] Computational methods - [ ] Biomolecular / infectious disease justification for this study - [ ] Figures - [ ] Data example (each class) - [ ] 3D Embeddings - [ ] Table of classification results / top features - [ ] Renderings of "exemplar" versions with features in play - [ ] Results - [ ] Discussion and Conclusions ### Todo Shannon: - [ ] Register on conference submission site Everyone: Next meeting: Monday, October 28 at 12:30pm October 21, 2024 --- :::info - **Time:** 12:30pm ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr - Alireza ::: Data on GACRC: `/project/spqlab/lab_shared/ornet/Confocal_3D_2024_ND2s/` (everything here should be current, as of 10/11) Amr's location: `/lustre2/work/spqlab/` ISBI Google doc: https://docs.google.com/document/d/10AFLGFnK6aaVcHIDjybOeCsXr-6eMONgLYW3u65wnlw/edit LaTeX paper: https://github.com/quinngroup/OrNet_ISBI_2025 Project timeline: https://github.com/orgs/quinngroup/projects/3/views/4 Initial analysis report: https://docs.google.com/document/d/1VI_CPuzqJxgOJ4jTQNzr6QFVsLJ-xA9fPdidHkmoSxg/edit ### Old Business **ISBI submission deadline extended to Oct 31** Controls were at 35C, may truly be showing a third and distinct phenotype - may need to use uninfected from other samples What is the z-stack voxel size? - Ask Jamie Barber - Amr, introduce Alireza to Jamie Is the new uninfected dataset ready? Information about temperature ranges of uninfected data in the google doc? ### New Business Checklist to completion: Analysis: - [ ] False positive / false negative / true positive / true negative exemplar analysis - [ ] 2-3 top features corresponding to classification - [ ] PCA embedding (debug / verify) - [ ] Linear feature ensemble between PCA and original feature space Experimental: - [ ] Final uninfected dataset - [ ] Post-mortem around exemplars - [ ] Discussion of features for classification Written: - [ ] Computational methods - [ ] Biomolecular / infectious disease justification for this study - [ ] Figures - [ ] Data example (each class) - [ ] 3D Embeddings - [ ] Table of classification results / top features - [ ] Renderings of "exemplar" versions with features in play - [ ] Results - [ ] Discussion and Conclusions Can we look at _degree of infection_? - sort cells by amount of fluorescence - could do this as a post-hoc step in examining results - what are the effects of more / less fluorescence on classification? ### Todo Amr: Alireza: - need help prioritizing computation - some incorrect reference descriptions in related works - need some help with annotation Shannon: - [ ] Register on conference submission site Everyone: Next meeting: Monday, October 21 at 12:30pm October 16, 2024 --- :::info - **Time:** 11:00am ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr - Alireza ::: Data on GACRC: `/project/spqlab/lab_shared/ornet/Confocal_3D_2024_ND2s/` (everything here should be current, as of 10/11) Amr's location: `/lustre2/work/spqlab/` ISBI Google doc: https://docs.google.com/document/d/10AFLGFnK6aaVcHIDjybOeCsXr-6eMONgLYW3u65wnlw/edit LaTeX paper: https://github.com/quinngroup/OrNet_ISBI_2025 Project timeline: https://github.com/orgs/quinngroup/projects/3/views/4 Initial analysis report: https://docs.google.com/document/d/1VI_CPuzqJxgOJ4jTQNzr6QFVsLJ-xA9fPdidHkmoSxg/edit ### Old Business **ISBI submission deadline extended to Oct 31** Controls were at 35C, may truly be showing a third and distinct phenotype - may need to use uninfected from other samples What is the z-stack voxel size? - Ask Jamie Barber - Amr, introduce Alireza to Jamie Amr, run another cell line - set up both 37C and 35C uninfected (controls) - staggered, so they can be imaged on different days Information about temperature ranges of uninfected data in the google doc? Introduced Alireza to Jamie Barber? ### New Business Amr has a new uninfected dataset, just hasn't been uploaded yet - check again later this week / early next week - just those that have been at 35C ### Todo Amr: - [ ] Need to include information about temperature ranges for uninfected data in the google doc - [ ] Additional uninfected data, given the time extension - [ ] Introduce Alireza to Jamie Barber Alireza: - [ ] Classification success / failure analysis - Look at data points that are consistently correctly and incorrectly classified; are there any interesting features about these? - [ ] Unsupervised domain analysis - PCA, UMAP, t-SNE - Color by wildtype/mutant, control, uninfected Shannon: - [ ] Register on conference submission site Everyone: - [ ] IEEE copyright forms - [ ] Info / paper content in google doc - [ ] Name and affiliation for paper Next meeting: Monday, October 21 at 12:30pm October 9, 2024 --- :::info - **Time:** 11:00am ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr - Alireza ::: Data on GACRC: `/project/spqlab/ornet/Confocal_3D_2024_ND2s/` (everything here should be current, as of 10/11) Amr's location: `/lustre2/work/spqlab/` ISBI Google doc: https://docs.google.com/document/d/10AFLGFnK6aaVcHIDjybOeCsXr-6eMONgLYW3u65wnlw/edit LaTeX paper: https://github.com/quinngroup/OrNet_ISBI_2025 Project timeline: https://github.com/orgs/quinngroup/projects/3/views/4 Initial analysis report: https://docs.google.com/document/d/1VI_CPuzqJxgOJ4jTQNzr6QFVsLJ-xA9fPdidHkmoSxg/edit ### Old Business First two are uninfected (should be about 7 cells) Ignore the last two - they don't seem to be infected - but there are bacteria in the media, so things could get confusing ### New Business **ISBI submission deadline extended to Oct 31** Controls were at 35C, may truly be showing a third and distinct phenotype - may need to use uninfected from other samples What is the z-stack voxel size? - Ask Jamie Barber - Amr, introduce Alireza to Jamie Amr, run another cell line - set up both 37C and 35C uninfected (controls) - staggered, so they can be imaged on different days ### Todo Amr: - [ ] Need to include information about temperature ranges for uninfected data in the google doc - [ ] Additional uninfected data, given the time extension - [ ] Introduce Alireza to Jamie Barber Alireza: - [ ] Classification success / failure analysis - Look at data points that are consistently correctly and incorrectly classified; are there any interesting features about these? - [ ] Unsupervised domain analysis - PCA, UMAP, t-SNE - Color by wildtype/mutant, control, uninfected Shannon: - [ ] Register on conference submission site Everyone: - [ ] IEEE copyright forms - [ ] Info / paper content in google doc - [ ] Name and affiliation for paper Next meeting: Wednesday October 16 at 11:00am October 7, 2024 --- :::info - **Time:** 12:30pm ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr ::: Data on GACRC: `/project/spqlab/ornet/Confocal_3D_2024_ND2s/` Amr's location: `/lustre2/work/spqlab/` ISBI Google doc: https://docs.google.com/document/d/10AFLGFnK6aaVcHIDjybOeCsXr-6eMONgLYW3u65wnlw/edit LaTeX paper: https://github.com/quinngroup/OrNet_ISBI_2025 Project timeline: https://github.com/orgs/quinngroup/projects/3/views/4 Initial analysis report: https://docs.google.com/document/d/1VI_CPuzqJxgOJ4jTQNzr6QFVsLJ-xA9fPdidHkmoSxg/edit ### Old Business Back catalog of uninfected data? ### New Business First two are uninfected (should be about 7 cells) Ignore the last two - they don't seem to be infected - but there are bacteria in the media, so things could get confusing With respect to the classifier: - if we don't see an effect globally, focus locally - look at mitochondria in areas of the cell specifically near the pathogen - unclear what to expect ### Todo Amr: - [ ] Need to include information about temperature ranges for uninfected data in the google doc Alireza: - [ ] Additional experiments and analysis Shannon: - [ ] Everyone: - [ ] IEEE copyright forms - [ ] Info / paper content in google doc - [ ] Name and affiliation for paper Next meeting: Wednesday October 9 at 11:00am September 30, 2024 --- :::info - **Time:** 12:30pm ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr ::: Data on GACRC: `/project/spqlab/ornet/new_nd2s/` Amr's location: `/lustre2/work/spqlab/` ISBI Google doc: https://docs.google.com/document/d/10AFLGFnK6aaVcHIDjybOeCsXr-6eMONgLYW3u65wnlw/edit LaTeX paper: https://github.com/quinngroup/OrNet_ISBI_2025 Project timeline: https://github.com/orgs/quinngroup/projects/3/views/4 ### Old Business From the previous meeting: - ISBI deadline is now October 11 (two weeks from this Friday). Initial Google doc already being edited by Russ - Has Amr added protocol information? - Amr has uploaded new data to GACRC (at /lustre location) - Additional replicates being generated (including controls with no pathogen) Start with September data, then move backward in time ### New Business 37C samples were contaminated :( 35C samples are still ok Back catalog of older uninfected data - Amr can go through it - Shannon will confirm with him Pay special attention to the fact the GFP channel will indicate what cells have been infected. - We need to focus our analysis on those cells - The other cells in the field of view will be uninfected, and therefore won't show the mitochondrial phenotypes we're looking for - need to focus on imaging cells that show pathogen actually invading the cells With respect to the classifier: - if we don't see an effect globally, focus locally - look at mitochondria in areas of the cell specifically near the pathogen - unclear what to expect ### Todo Amr: - [ ] Finish additional replicates - [ ] Catalog old data Alireza: - [ ] Segment individual cells - [ ] Ask Vet Med team for assistance with any ambiguities - [ ] Calculate feature vectors per cell for final frames of all videos (these last frames should show the greatest deltas in the mitochondrial morphologies) - [ ] Train a linear classifier with these feature vectors and see if we can differentiate control/mutant from wildtype (a linear classifier will be more easily interpreted, in particular telling us which features are most informative to classification) Next meeting: Monday October 7 at 12:30 September 24, 2024 --- :::info - **Time:** 3:00pm ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr - Alireza ::: Data on GACRC: `/project/spqlab/ornet/new_nd2s/` Amr's location: `/lustre2/work/spqlab/` ISBI Google doc: https://docs.google.com/document/d/10AFLGFnK6aaVcHIDjybOeCsXr-6eMONgLYW3u65wnlw/edit LaTeX paper: https://github.com/quinngroup/OrNet_ISBI_2025 Project timeline: https://github.com/orgs/quinngroup/projects/3/views/4 ### Old Business From the previous meeting: - ISBI deadline is now October 11 (two weeks from this Friday). Initial Google doc already being edited by Russ - Has Amr added protocol information? - Amr has uploaded new data to GACRC (at /lustre location) - Additional replicates being generated (including controls with no pathogen) ### New Business Alireza has questions about existing data - Momentary dimming in brightness is some kind of an artifact that we need to find a way to deal with - `v` channel contains images of different wells from the same experiment (though they are labeled to indicate wildtype (wt), mutant (esat6 ko), or control (no pathogen)) Russ: potential later effects of ESAT-6 KO infection - need to focus on imaging cells that show pathogen actually invading the cells Some older videos show cross-reaction between both mitochondria and tagged marinum under the green laser (i.e. the green channel shows both marinum and mitochondria) - Seems like some minor autoscaling of LUTs will help bring down the relative background brightness Start with September data, then move backward in time With respect to the classifier: - if we don't see an effect globally, focus locally - look at mitochondria in areas of the cell specifically near the pathogen - unclear what to expect ### Todo Amr: - [ ] Finish additional replicates Alireza: - [ ] Segment individual cells - [ ] Ask Vet Med team for assistance with any ambiguities - [ ] Calculate feature vectors per cell for final frames of all videos (these last frames should show the greatest deltas in the mitochondrial morphologies) - [ ] Train a linear classifier with these feature vectors and see if we can differentiate control/mutant from wildtype (a linear classifier will be more easily interpreted, in particular telling us which features are most informative to classification) Next meeting: Monday September 30 between 12 and 2 September 16, 2024 --- :::info - **Time:** 9:00am ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr ::: Data on GACRC: `/project/spqlab/ornet/new_nd2s/` Amr's location: `/lustre2/work/spqlab/` ISBI Google doc: https://docs.google.com/document/d/10AFLGFnK6aaVcHIDjybOeCsXr-6eMONgLYW3u65wnlw/edit LaTeX paper: https://github.com/quinngroup/OrNet_ISBI_2025 Project timeline: https://github.com/orgs/quinngroup/projects/3/views/4 ### Old Business From the previous meeting: - Russ has started supplying text on the SciPy document - We need a new google doc specifically for ISBI - Have we seen any evidence of bacteria dividing? - Amr has uploaded new data to GACRC (at /lustre location) ### New Business - ISBI 2025 deadline is now **Oct 11**! (a two-week extension) - We'll use that time to generate some additional replicates - Including controls with no pathogen ### Todo Shannon - [ ] Create a new google doc, fill in Russ' edits, and send out to everyone Alireza - [ ] Complete the GACRC training and get access to data - [ ] Start on analysis - [ ] Get Amr protocols and add to document Amr - [ ] Upload more data (replicates?) to GACRC - [ ] Dump text of experimental protocols somewhere so Alireza can add to ISBI draft - [ ] Evidence of bacteria dividing? Next meeting: Wednesday, September 25 at 11am September 9, 2024 --- :::info - **Time:** 1:00pm ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr - Alireza ::: Data on GACRC: `/project/spqlab/ornet/new_nd2s/` - Back up and running! Amr's location: `/lustre2/work/spqlab/` GitHub repos: - https://github.com/quinngroup/mitometer-conversion - https://github.com/quinngroup/ornet ### Old Business From the previous meeting: - Got some *M. marinum* mutants! They're here, they're growing (while we wait for the scope to be repaired) - no ESAT-6 knockouts - have asked for mutants in larger genes flanking ESAT-6 and its partner protein - provided insertion mutants in genes on either side - need to do some genetics: remove the drug marker, put in GFP plasmid GACRC was down for months - FINALLY restored the `/project` mount (which our `spqlab` was on) - Should work again; put and get all data there ### New Business [ISBI 2025](https://biomedicalimaging.org/2025/) Submission deadline: Sept 27 Wildtype and knockout infections! - Is there a difference in the movement/shape/quantity of mitochondria between mutant (knockout) bacteria vs wildtype? - May need to look at green signal to ascertain whether the bacteria actually managed to invade and get inside the cell - Look for hyperlocalized effects on mitochondria around the bacteria, as opposed to global changes ### Todo Shannon - [ ] Amr and Alireza still can't access GACRC - [ ] Amr will upload to "old" spot; I'll move it over, while waiting for GACRC to give them access - [ ] Alireza needs to complete training - [x] Read through Alireza's notes on Slack - [x] Find SciPy abstract and send to Amr Alireza - [x] Complete the GACRC training and get access to data - [x] Start on analysis - [ ] Get Amr protocols and add to document Amr - [ ] Upload more data (replicates?) to GACRC - [ ] Dump text of experimental protocols somewhere so Alireza can add to ISBI draft - [ ] Evidence of bacteria dividing? Next meeting: Monday, September 16 at 9am March 27, 2024 --- :::info - **Time:** 3:00pm ET - **Agenda** 1. Project status / Old business 2. New business 3. Action items - **Participants:** - Drs. Quinn - Dr. Karls - Amr ::: Data on GACRC: `~/spqlab/lab_shared/ornet` GitHub repos: - https://github.com/quinngroup/mitometer-conversion - https://github.com/quinngroup/ornet ### Old Business From the previous meeting: - "justification" for final ~$4k of promised SPA money submitted a couple weeks ago - generate mutant data - SciPy 2024 abstract submitted! Reviewer feedback deadline is March 31, so we should hear about initial acceptance soon - ASM Microbe with an abstract submission deadline of **April 16 at NOON** (opens April 2) - https://asm.org/Events/ASM-Microbe/Present - GACRC data transfer problems? - Amr used WinSCP to upload videos, but they weren't showing up. Asked for screenshots to show process - 3D imaging - red is mitochondria; green is GFP bacteria (100 MOI) - fix was to image one well at a time - will definitely need nd2 files (the AVIs flatten the 3D information) - Bacterial infection - Need Jamie's help; something is not working in the protocol - **Mutant protocol is stuck at the plasmid level** - Have to get plasmid integrated into the chromosome - Last cell transformation with ESAT-6 knockout plasmid: ~2 months ago (plates are still in the incubator, but none are showing any changes to indicate successful transformation) - **Need 1000 transformants to get 1 recombination event** ### New Business Green laser has been down since our last (Feb) discussion Got some *M. marinum* mutants! They're here, they're growing (while we wait for the scope to be repaired) - no ESAT-6 knockouts - have asked for mutants in larger genes flanking ESAT-6 and its partner protein - provided insertion mutants in genes on either side - need to do some genetics: remove the drug marker, put in GFP plasmid Let's start working on the ASM Microbe abstract (due April 16) (June 13-17) - start with SciPy abstract ([here](https://zoom.us/j/91907322975?pwd=Nk83cWx0MGV0dXlmN3hUaEcxWmdIdz09)) and this is the [conference](https://www.scipy2024.scipy.org/) (July 8-14, with the main conference July 10-12) ### Todo - [ ] Amr: check with GACRC about location of uploaded data - [ ] Reach out to Shannon after hearing from them (include screenshots if possible) - [ ] Start on ASM Microbe abstract (just under 3 weeks; can we use LLO related analysis from previous work?) - [ ] Send out an update / proposal on work we could do - [ ] Reach out to Ali's old thesis advisor (Vini)? - [ ] Dig through old LLO/Mdivi data - [ ] Contact again by email by April 3 - [ ] Next meeting: Wednesday, April 17, 3pm February 21, 2024 --- :::info - **Time:** 3:00pm ET - **Agenda** 1. Project status 2. Goals and timeline 3. Other project updates - **Participants:** - ::: Data on GACRC: `~/spqlab/lab_shared/ornet` GitHub repos: - https://github.com/quinngroup/mitometer-conversion - https://github.com/quinngroup/ornet ### Old Business Apparently I'm 0.8 months short of budgeted effort and practically the entirety of what remains of the budget has to go to that - Should we ask SPA for more money? **Yes** - Something about "if this isn't enough you can ask for more"--need to track this down - Check out "Projected Shortfall" tab Janelia 4D abstract wasn't accepted :( Issues with 3D imaging - ran into problems with recording time series (blurry images) - Changes to the imaging protocol? ### New Business Amr used WinSCP to upload videos to `spqlab` on GACRC - however they are not appearing in the root folder 3D imaging - red is mitochondria; green is GFP bacteria (100 MOI) - fix was to image one well at a time - will definitely need nd2 files (the AVIs flatten the 3D information) Bacterial infection - Need Jamie's help; something is not working in the protocol - **Mutant protocol is stuck at the plasmid level** - Have to get plasmid integrated into the chromosome - Last cell transformation with ESAT-6 knockout plasmid: ~2 months ago (plates are still in the incubator, but none are showing any changes to indicate successful transformation) - **Need 1000 transformants to get 1 recombination event** SciPy 2024 Conference: abstract submission deadline **Feb 27.** - If abstract is accepted, we're invited to submit a full paper (due May 31) - Paper goes through peer review - Accepted paper will result in a poster invitation - https://cfp.scipy.org/2024/cfp ASM Microbe with an abstract submission deadline of **April 16 at NOON** (opens April 2) - https://asm.org/Events/ASM-Microbe/Present ### Todo - [x] Come up with a budget / justification to submit to SPA for extra money - look into what language would be required - [x] Start with Projected Shortfall - [ ] Generate the mutant data - [x] Draft an abstract for SciPy - [x] Send abstract to Amr + everyone (start with Janelia) - [x] Next meeting: **March 20 at 3pm** (send out invites) November 8, 2023 --- :::info - **Time:** 4:00pm ET - **Agenda** 1. Project status 2. Goals and timeline 3. Other project updates - **Participants:** - Drs. Quinn & Quinn - Dr. Karls - Alireza - Amr ::: Data on GACRC: `~/spqlab/lab_shared/ornet` GitHub repos: - https://github.com/quinngroup/mitometer-conversion - https://github.com/quinngroup/ornet Janelia 4D Cellular Physiology abstract **submitted!** 3D imaging - running into problems with recording time series: getting blurry images on every time frame after the first - Hopefully will have the protocol nailed down by our next meeting for generating 3D data - In the meantime, work with 2D data Working on infection protocols / plasmids - *M. marinum* is our go-to Current R21 budget is at **$13.6K** at end of CY 2023. - Should we go back to SPA and ask for more money? Next meeting: **Wednesday, December 13, 3pm** October 11, 2023 --- :::info - **Time:** 3:00pm ET - **Agenda** 1. Project status 2. Goals and timeline 3. Other project updates - **Participants:** - Drs. Quinn & Quinn - Dr. Karls - Amr ::: Data on GACRC: `~/spqlab/lab_shared/ornet` GitHub repos: - https://github.com/quinngroup/mitometer-conversion - https://github.com/quinngroup/ornet 3D images! ![](https://hackmd.io/_uploads/S1ZwX_V-a.png) - Red are fluorescently-tagged mitochondria in a single A549 cell (just can't see the cell wall) - Green are fluorescently-tagged bacteria surrounding the cell **Advantages** - 3D! - Still have time series - Lots more detail - Photobleaching has improved dramatically **Disadvantages** - Some issues around being able to move the lens between images (as it is an oil-immersion lens) -- results in not being able to image multiple wells over a single time course Could we do some superresolution? - we can do high-throughput 40x imaging - use large 40x data + superresolution to get to 100x oil immersion dataset Bacteria don't seem to be infecting cells, for some reason - bacteria clumped up and grouped around the cell - but didn't invade, why? Goals for data generation: - 5-10 examples of uninfected cells Computational goals: - **need a tracklet module**: we need to replicate the tracking portion of Mitometer so we can generate tracks on our own. - tracking in 3D (toxo work) - some approaches of parameterizing / embedding phenotype appearance Next meeting: **Wednesday, November 8, 4pm** June 22, 2023 --- :::info - **Date:** June 22, 2023, 3:00pm ET - **Agenda** 1. Project status 2. Goals and timeline 3. Other project updates - **Participants:** - Drs. Quinn & Quinn - Dr. Karls - Alireza - Rachel - Amr ::: Project Status --- - NCE approved - Project end: June 30, 2024 - Have tracking code ready for testing (still trying to get some parameters fixed) https://github.com/quinngroup/Mitometer - Will hopefully have some tracks ready soon! - Data has a resting place on GACRC: `~/spqlab/lab_shared/ornet` - Some observations - MOI 10, 100, 1000 - 10 looks almost wildtype - 100 sees a progression - 1000 looks like Llo - What is the GFP? looks like it's outside the cell - What is "TD"? (some kind of contrast, since "noTD" looks fine) - x10 probably not enough - x60 more what we want (could maybe do x100) - Filename questions - geneticin? - DSRedMito "with green"? - File formats: .tif, .mp4, .nd2, .avi - We prefer .avi, but .tif will also work - New GFP plasmid! - old GFP plasmid was problematic, hence lack of green signal - Created the vector for knocking out the ESAT-6 gene - Need to get it into the genome of Mtb - Verify it got in, knocked out the gene, got out - Going to start generating z-stack (3D!!) data as well - Will dramatically help with tracking precision (bring in the 3D toxo tracking strategies) Goals and Timeline --- Rachel's defense (summer 2023) Amr's GACRC training is TOMORROW Action items --- - **QRG** - [ ] Find some symposia / conference where presenting this would be appropriate - [ ] Keystone tuberculosis conference in January - [ ] Run tracking on all the data - [ ] Clustering / parameterization of tracks - [ ] Detectable differences based on the following :arrow_down: - [ ] Go through Amr's uploaded videos - [ ] What parameter combinations are optimal (for us)? - [x] **What MOI levels (infection concentration) are "optimal" for us?** - [x] Provide this feedback to Amr - **ID** - [ ] More A549 wt data - [ ] By themselves - [ ] Infected - [ ] Negative controls - [ ] Smegmatis tags - [ ] z stacks ## Notes <!-- Other important details discussed during the meeting can be entered here. --> --- Internal Meeting Minutes === June 21, 2023 --- :::info - **Date:** June 21, 2023, 3:00pm ET - **Agenda** 1. Project updates 2. Goals and timeline - **Participants:** - Alireza - Rachel - Shannon ::: Project Updates --- **Alireza** - Code complete for the particle tracking https://github.com/quinngroup/Mitometer - Some performance issues - Wondering about the role of some of the hyperparameters (e.g., MicronsPerPixel) in the track calculation - Are these necessary? - Run the tracking against the data **once** and then operate on the tracks thereafter **Rachel** - interactive plotting - bokeh, plotly Goals and Timeline --- Rachel: still planning to defend in summer 2023 Action items --- **Alireza** - [ ] Run Mitometer on all the videos we have - [ ] Use ffmpeg to convert AVIs to TIFFs - [ ] Use `scipy.io` utility to read Matlab output files https://docs.scipy.org/doc/scipy/reference/io.html#matlab-files - [ ] Start designing the post-processing / analysis side of things (i.e., how we analyze the tracks that come out of Mitometer) **Rachel** - [ ] Push exploratory code out to a repo - [ ] Start digging into initial results (particularly Sparse PCA) to try and explain axes - [ ] Rollover / hover effects on data points as a way to improve explainability - [ ] Mix in some other embedding methods (UMAP, tSNE, NMF) Notes --- <!-- Other important details discussed during the meeting can be entered here. --> --- OrNet All-Hands Meeting Minutes === April 27, 2023 --- :::info - **Date:** April 27, 2023, 3:00pm ET - **Agenda** 1. Project status 2. Goals and timeline 3. Other project updates - **Participants:** - Drs. Quinn & Quinn - Dr. Karls - Alireza - Rachel - Amr ::: Project Status --- **Project State** - Comparison of wildtype / virulent *M. Marinum* to knockout / mutant version - Key protein: ESAT-6 - Absent in the knockout / mutant version - More A549 cells sitting around, not getting infected - Want to have some control infections with *M. tuberculosis* - Some approved BSL-2 strains for generating control data - Two versions: one can get out of the phagosome to impact mitochondrial morphology, the other can't - Should provide a means for comparison - Fixed cells? Maybe - Would be tough to compare between them, with no contextual dynamics Goals and Timeline --- ID: Generating more data, additional protocols for infections - A549 wt, by themselves - A549 wt, infection NCE submitted - If approved, project extended to June 30, 2024 Next All-Hands? **Thursday, May 25, 3pm** Action items --- - **QRG** - [ ] Go through Amr's uploaded videos - [ ] What do the filenames mean? - [ ] What parameter combinations are optimal (for us)? - [ ] **What MOI levels (infection concentration) are "optimal" for us?** - [ ] Provide this feedback to Amr - [x] Change / Re- setup FTP server for more data - [x] Check out GACRC - https://wiki.gacrc.uga.edu/wiki/Disk_Storage#Project_file_system - https://uga.teamdynamix.com/TDClient/Requests/ServiceDet?ID=25839 - **ID** - [ ] More A549 wt data - [ ] By themselves - [ ] Infected - [ ] Negative controls - [ ] Smegmatis tags ## Notes <!-- Other important details discussed during the meeting can be entered here. --> --- OrNet Internal Meeting Minutes === April 27, 2023 --- :::info - **Date:** April 27, 2023, 2:00pm ET - **Agenda** 1. Project updates 2. Goals and timeline - **Participants:** - Alireza - Rachel - Shannon ::: Project Updates --- **Alireza** - Created a GitHub repo for Mitometer https://github.com/quinngroup/Mitometer - JUST the particle tracker - What about just running the Matlab tracker on ALL our videos and dumping the data to read into Python for post-processing? - AVI to TIFF conversion - Currently using a hand-made OpenCV script to do the conversion - What about ffmpeg? **Rachel** - Trying to train a model to learn fission vs fusion - Sparse PCA, LDA - Problems with test set being very small (is this something we can talk to Amr about?) - Some shapes! SUPER COOL! - Code? Goals and Timeline --- Rachel: still planning to defend in summer 2023 Action items --- **Shannon** - [x] Set up a new FTP server for Amr to upload data - [x] Set up a summer 2023 meeting schedule **Alireza** - [ ] Run Mitometer on all the videos we have - [ ] Use ffmpeg to convert AVIs to TIFFs - [ ] Use `scipy.io` utility to read Matlab output files https://docs.scipy.org/doc/scipy/reference/io.html#matlab-files - [ ] Start designing the post-processing / analysis side of things (i.e., how we analyze the tracks that come out of Mitometer) **Rachel** - [ ] Push exploratory code out to a repo - [ ] Start digging into initial results (particularly Sparse PCA) to try and explain axes - [ ] Rollover / hover effects on data points as a way to improve explainability - [ ] Mix in some other embedding methods (UMAP, tSNE, NMF) Notes --- <!-- Other important details discussed during the meeting can be entered here. --> Location of Amr's new OrNet data (**note: this is a BACKUP location for the time being**): - `rocinante.ltg.cs.uga.edu` - `/space/atlantis/ftp/ornet/` Next update: TBD Next meeting: TBD **Next All-Hands**: --- March 23, 2023 --- :::info - **Date:** March 23, 2023, 2:00pm ET - **Agenda** 1. Project updates 2. Goals and timeline - **Participants:** - Alireza - Rachel - Shannon ::: Project Updates --- **Alireza** - script-ify Mitometer **Rachel** - pretty much ready to run the first round of experiments - need to figure out how Mitometer calculates fission/fusion Goals and Timeline --- ~~Rachel's thesis due March 25 to Graduate School~~ Moved graduation to summer, so timeline for thesis deadline has changed Action items --- - [x] Alireza will strip out part of Mitometer that generates tracks and "script-ify" it to work with new data - [ ] Create a repo for this work and post tickets - [ ] Still waiting on status of Logan from Piotr Notes --- <!-- Other important details discussed during the meeting can be entered here. --> Location of Amr's new OrNet data: - `atlantis.cs.uga.edu` - `/srv/ftp/ornet/` Next update: Thursday, March 30 Next meeting: Thursday, March 30 **Next All-hands**: Thursday, March 30 @ 3pm --- February 16, 2023 --- :::info - **Date:** February 16, 2023, 2:00pm ET - **Agenda** 1. Project updates 2. Goals and timeline - **Participants:** - Meekail - Alireza - Rachel - Shannon ::: Project Updates --- **Alireza** - explored converting the matlab code to python 2.7 - still need to convert to python 3 - tried chatgpt conversion and actually seemed ok - mitometer itself seems pretty cool - **at the very least, its method for identifying tracks would be useful** **Rachel** - need more checklists (GitHub issues?) - https://github.com/rmattson1008/ornet - PyTorch Lightning may have been overkill - Migrate to regular PyTorch? - `seed_everything` - Use previous publications as preliminary / background! Goals and Timeline --- Rachel's thesis due March 25 to Graduate School Action items --- - [ ] Reach out to Amr to see if he's done uploading data, and what the uploaded data actually are (what kinds of videos? different types? what are the dimensions?) - [ ] What are microns/pixel (or microscope resolution)? - [ ] What is the time lapse between frames? - [ ] How does mitometer identify tracks? - [ ] **Alireza**: create a repo and start posting tickets for updates Notes --- <!-- Other important details discussed during the meeting can be entered here. --> Next update: Thursday, February 23 Next meeting: Thursday, March 2 --- OrNet All-Hands Meeting Minutes, Jan 26 2023 === :::info - **Date:** January 26, 2023, 2:30pm ET - **Agenda** 1. Project status - Old/New personnel - Current project state 2. Goals and timeline - This is year 2 - Remaining goals 3. Other project updates - **Participants:** - Drs. Quinn & Quinn - Dr. Karls - Meekail - Alireza - Rachel - Amr ::: Project Updates --- **Personnel** Shannon Quinn's group: - Postdoc departed Dec 31 - Rachel will be graduating this semester with her MS in AI - Thesis focus: focus on fissioning and fusioning events - bit of a mixture model on the video embeddings; allows for ratios of fission/fusion events - Alireza joining to help test published methods - Mitometer: https://github.com/aelefebv/Mitometer - Functionality - Port to Python? - Meekail available as consultant - Some ideas on anomaly detection **Project State** - Comparison of wildtype / virulent *M. Marinum* to knockout / mutant version - Key protein: ESAT-6 - Absent in the knockout / mutant version - **Some challenges with making the knockouts** - Went with a published system but didn't seem to be stable in *e. coli* (problems with the method? with the repository information?) - Went back to an older method, but requires much larger plasmid, which reduces frequency with which plasmid gets into bacteria, hence reducing recombination events (and reducing overall population of mutants) - Amr has been working on this for awhile - Want to have some control infections with *M. tuberculosis* - Some approved BSL-2 strains for generating control data - Two versions: one can get out of the phagosome to impact mitochondrial morphology, the other can't - Should provide a means for comparison Goals and Timeline --- Spring 2023 and Summer 2023 (grant technically ends 6/30) Action items --- - [x] Set up an FTP link to logan / atlantis - Talk to Meekail about getting this set up - Send the link to EVERYONE but especially Amr to upload the data that was put on WeTransfer back in July 2022 (it seems to have gone missing) - [ ] Look at the data Amr generated (once we've received it) from last year, and ask: can we tell the difference between infection and no? - [ ] Schedule next meeting in two months (end of March / early April) ## Notes <!-- Other important details discussed during the meeting can be entered here. --> --- OrNet Meeting Minutes, Dec 19 2022 === :::info - **Date:** December 19, 2022, 10:00am ET - **Agenda** 1. Project status - Data / Experiments - Personnel 2. What are our goals? - This is year 2 3. Other project updates - **Participants:** - Shannon - Russ - Fred - **Host:** Shannon ::: Project Status --- - Postdoc is leaving Dec 31; Rachel defending Master's thesis either May or August 2023 on this topic - Amr has been funded for 1.5 years now - Current cloning method hasn't been working - Old method for mutant screening is laborious but at least it works - will pick this up, with focus on ESAT-6 (since we know this knockout has an effect) Goals --- - We need a positive control, and a negative control - M. marinum with GFP plasmid ESAT-6 knockout is our positive - We do have an Mtb strain _lacking_ ESAT-6 (won't be able to replicate inside host cells, but will at least be able to invade; BSL-2) - not a perfect negative control, but still a potential option - Push this angle while Amr continues working on new deletion plasmids Other project updates --- - Amr will be teaching Spring 2023 - **Shannon** - [ ] Check in on budget status with Franklin folks - [ ] See if Meekail wants to use remaining postdoc budget to develop some analysis (work with Rachel, too?) - [ ] Schedule an all-hands meeting for first week of February 2023 ## Notes <!-- Other important details discussed during the meeting can be entered here. --> --- :::info - **Date:** May 23, 2022, 11:00am ET - **Agenda** 1. Follow-up from Feb 11 (!) and March 25 meetings 2. SciPy 3. Other project updates - **Participants:** - Shannon - Meekail - Rachel - Victoria - Delaney - **Host:** Shannon ::: :books: Follow-up from Feb 11 --- - Take a fixed ROI, apply a CNN to it, and compare first frame to last frame - next tier up would be to incorporate global information using CNN (e.g., ResNet model) across entire image - shortcut between the two using a strict ROI on every frame or every other frame, and running some temporal model on the series of path vectors - Texture tesselation! - Would be useful in both OrNet and cilia projects - What methods are there for this? - Baseline CNN model with tiered improvements up to LSTM - testing against a suite of simple classifiers (KNN, linear SVM, one-layer perceptron) - Victoria & Delaney - Focused on the SciPy paper - Introduction, lit review - Methods? (haven't gotten to that yet) - Did something ROI related - Latest repo work: https://github.com/Micky774/ornet - Latest draft paper: https://docs.google.com/document/d/1m2eBVjEeT56eJP525CsXLOUHbefDvitjqabCnLkRsoA/edit - Rachel - Getting the model to train - How robust are the results? - Potential new papers to read - https://www.biorxiv.org/content/10.1101/2021.12.03.470962v1 (linked code is questionable, but methods and general task of identifying and segmenting organelles could be useful) - https://www.sciencedirect.com/science/article/pii/S2589004220307938 MitoSegNet: [supposedly] a fully trained deep network for :snake: SciPy papers --- - Paper draft deadline **May 27** - What are the goals that need to be done? - Sanitizing results - Statistical viability for train/test splits - TensorBoard logging - Different initialization strategies :dart: Other project updates --- ## Goals for this week :closed_book: Tasks -- ==Importance== (1 *most* - 5 *least*) / Task / **Estimate** (# of hours) ## Notes <!-- Other important details discussed during the meeting can be entered here. --> - Neelima's / Marcus' GMM investigation seemed to indicate that they weren't any more expressive than standard linear models - Neelima's highest classification accuracy was Random Forests (RFs) on intermediate graph statistics - Rachel: take a fixed ROI, apply a CNN to it, and compare first frame to last frame - next tier up would be to incorporate global information using CNN (e.g., ResNet model) across entire image - shortcut between the two using a strict ROI on every frame or every other frame, and running some temporal model on the series of path vectors - Texture tesselation! - Would be useful in both OrNet and cilia projects - What methods are there for this? --- :::info - **Date:** March 25, 2022, 1:00pm ET - **Zoom Link:** [zoom](https://zoom.us/j/93188047818?pwd=N09pdnBPWGtEUXphTDhzbkVQU1VkZz09) - **Agenda** 1. [Introduction and Project Overview](https://hackmd.io/miXLBO68TnGcve1-KBRSXQ#1-Introduction-and-Project-Overview) 2. [Laboratory updates](https://hackmd.io/miXLBO68TnGcve1-KBRSXQ#2-Laboratory-updates) 3. [Computational updates](https://hackmd.io/miXLBO68TnGcve1-KBRSXQ#3-Computational-updates) 4. [Goals](https://hackmd.io/miXLBO68TnGcve1-KBRSXQ#4-Goals) 5. [Next meeting](https://hackmd.io/miXLBO68TnGcve1-KBRSXQ#5-Next-meeting) - **Participants:** - Prof. Fred Quinn - Prof. Russell Karls - Prof. Shannon Quinn - Dr. Blessing Ojeme - Meekail - Rachel - Victoria - Marcus - Delaney - Amr - **Host:** Fred / Russ / Shannon ::: ## 1: :wave: Introduction and Project Overview Shannon - rando Marcus - PhD candidate in CS Meekail - PhD student in CS Victoria - undergrad in bio engineering Delaney - undergrad in math & data science Rachel - undergrad in cog sci Russ - research scientist and molecular biologist Fred - professor and dept head in infectious diseases Amr - 2nd year graduate student ## 2: :microscope: Laboratory updates Thoughts here. ### a. Development of *Mycobacterium marinum* mutants ![](https://i.imgur.com/R9NvrGT.png) Process enables recombination and, therefore, replication with genetic modifications. Currently cloning regions upstream and downstream of target region for inserting desired modifications. ESAT-6 is a known gene in Mtb that is required for the observed phenomenon in mitochondria. Knocking it out gives the punctate mitochondrial phenotype we're familiar with. ![](https://i.imgur.com/TshjtvK.png) More mutants! Yay! Mix of known and unknonwn mutants. ### b. Culture and confirmation of A549 alveolar epithelial cells expressing DsRed-mitochondria ![](https://i.imgur.com/QWtUo7e.jpg) (10x mag) Same cell line that Aly used. ### c. Generating and screening GFP-expressing *Mycobacterium marinum* ![](https://i.imgur.com/OTsV2Xw.jpg) Oooh! Some bacteria have lost the ability to fluoresce. - expression of GFP seems uneven in mutants - some grow fast but do not fluoresce, while others grow slowly but fluoresce brightly ### d. Pilot A549-DsRed/*M. marinum*-GFP infections ![](https://i.imgur.com/jdmLi9k.png) Cool! Apparently glass vs plastic slides are not compatible with higher-magnification objectives ## 3: :desktop_computer: Computational updates ### a. Network modeling progress Andrew's work (pre-2020) was all about creating the graph representation. Problem of multiple cells. Used modeling strategy called [Gaussian Mixture Model](https://en.wikipedia.org/wiki/Mixture_model#Gaussian_mixture_model) to create graph structure (nodes/edges). ![](https://i.imgur.com/L0dzAQC.png) Spectral graph theory - transform graph structures into a matrix representation called the [graph Laplacian](https://en.wikipedia.org/wiki/Laplacian_matrix) - decompose laplacians via eigen-decomposition into eigenvalues and eigenvectors - did this for every sample / image in a stack and tracked the changes in these values over time - **look for shifts in eigenvalues over time** and hope? that those shifts correlate with phenotypic changes ![](https://i.imgur.com/15tHqyL.png) Also conducted testing on how good our graph representation of the image data was. - sort of a sanity check on our work - can an off-the-shelf statistical model pick up on differences between experimental conditions based solely on our graphical representation? - if so, our graph representation is capturing important information - if not, then our representation is bad (and should feel bad) There are going to be more *regional* effects due to infection - LLO is something of a sledgehammer: all of the mitochondria are affected by it - Infection by pathogen will be more localized in a single cell - Eventually this will encompass the whole cell, but initially it will be a localized phenomenon ### b. Automated segmentation *cut for time* ## 4: :dart: Goals - Two abstract submissions to [SciPy 2022](https://www.scipy2022.scipy.org/) (one on modeling, one on segmentation) ![](https://i.imgur.com/T5CDQG0.jpg) EM image of A549 cells - A: infected with wildtype Mtb - B: infected with mutant strain Mtb (3251-C knockout) **Morphology of mitochondria is indicative of cellular outcome** - necrosis vs apoptosis - a lot of bacterial genes are involved in this process - **if we can prevent this from happening and keep the mitochondria healthy, we could use the corresponding genes as drug/vaccine targets** ## 5: :speaking_head_in_silhouette: Next meeting - [ ] Thursday, April 7 at 1pm - [ ] Friday, April 8 at 11am - [ ] Friday, April 8 at 1pm - [ ] Thursday, April 14 at 1pm - [ ] Friday, April 15 at 11am - [ ] Friday, April 15 at 1pm :notebook: Meeting Notes --- Notes are all inline above but feel free to put extra thoughts here! --- ###### tags: `Templates` `Meeting` :::info - **Date:** Feb 11, 2022, 11:00am ET - **Agenda** 1. Follow-up from Jan 28 meeting 2. SciPy abstracts 3. Other project updates - **Participants:** - Shannon - Blessing - Meekail - Rachel - Victoria - Marcus - **Host:** Shannon / Blessing ::: :books: Follow-up from Jan 28 --- - Set meeting time for Victoria, Delaney, Rachel, Blessing, and Meekail - needs to be rescheduled - Two planned SciPy abstracts - baseline model CNN - eliminate inductive biases (i.e., social network) and see how they work - SciPy-based reimplementation of Mitometer (more details to come) :snake: SciPy abstracts --- - Due TONIGHT - Mitometer paper ([nature url](https://www.nature.com/articles/s41592-021-01234-z)) - PDF: https://www.nature.com/articles/s41592-021-01234-z.pdf?origin=ppub - GitHub repo https://github.com/aelefebv/Mitometer - Baseline CNN model with tiered improvements up to LSTM - https://docs.google.com/document/d/1axKvxsQB3OG5ki7WxL19CJicYKodFpcIq0uhbn7afQA/edit?usp=sharing - testing against a suite of simple classifiers (KNN, linear SVM, one-layer perceptron) :dart: Other project updates --- - Changing `master` branch to `main` on OrNet repo - Marcus: - writing, writing, writing - running additional experiments - rethinking object detection and tracking - density-based clustering [HDBSCAN](https://hdbscan.readthedocs.io/en/latest/how_hdbscan_works.html) - [Hungarian algorithm](https://en.wikipedia.org/wiki/Hungarian_algorithm) for object tracking - details of videos would indicate presence of non-uniform mitochondrial phenotypes - heterogeneous *distributions* of multiple mitochondrial phenotype within a single cell! zomg - could potentially model classification as a multi-label task, e.g. K-many binary classifiers while only one gets trained at a time - ![](https://i.imgur.com/7VN3B4A.png) - Only concern with HDBSCAN is the hyperparameters - How do they translate to the biology? - Why is one set of hyperparameters "better" than another? ## Goals for this week - Submit SciPy abstracts! - Marcus: use [Ridler-Calvard to perform histogram-based thresholding](https://scikit-image.org/docs/stable/api/skimage.filters.html#threshold-isodata) to convert image data for use by HDBSCAN - [Connected component analysis](https://scikit-image.org/docs/stable/api/skimage.measure.html#label) to isolate specific regions - Reduce each _region_ to a single _gaussian_ parameterized by the centroid and covariance (should be available through [regionprops](https://scikit-image.org/docs/dev/api/skimage.measure.html#regionprops)) - Generate a number of samples in *euclidean space* from the gaussian based on the intensity represented by the original connected component - The point is: renormalizing fluorescence in a way that HDBSCAN is invariant to - **Work with Meekail on this** :closed_book: Tasks -- ==Importance== (1 *most* - 5 *least*) / Task / **Estimate** (# of hours) - [ ] ==1== Reschedule weekly OrNet meeting (conflicts with scikit-learn triage meeting) - [ ] ==1== Finish SciPy abstract on baseline methods - [ ] ==5== Find a texture tesselation method - [ ] Meekail: Reach out to Dr. Bhandarkar - [ ] Shannon: Reach out to Chakra - [ ] ==1== Change `master` branch to `main` branch ## Notes <!-- Other important details discussed during the meeting can be entered here. --> - Neelima's / Marcus' GMM investigation seemed to indicate that they weren't any more expressive than standard linear models - Neelima's highest classification accuracy was Random Forests (RFs) on intermediate graph statistics - Rachel: take a fixed ROI, apply a CNN to it, and compare first frame to last frame - next tier up would be to incorporate global information using CNN (e.g., ResNet model) across entire image - shortcut between the two using a strict ROI on every frame or every other frame, and running some temporal model on the series of path vectors - Texture tesselation! - Would be useful in both OrNet and cilia projects - What methods are there for this? --- :::info - **Date:** Jan 28, 2022, 11:00am ET - **Agenda** 1. Catch-up from fall 2021 2. Plans for spring 2022 - **Participants:** - Shannon - Blessing - Meekail - Rachel - **Host:** Shannon / Blessing ::: :books: Fall 2021 Review --- - Backlog - Automated mitochondrial segmentation https://www.nature.com/articles/s41592-021-01234-z :dart: Spring 2022 Goals --- - Marcus: Graduate! ## Goals for this week - Rachel: note single cell data that failed segmentation :closed_book: Tasks -- ==Importance== (1 *most* - 5 *least*) / Task / **Estimate** (# of hours) - [ ] ==3== Read paper **2** ## Notes <!-- Other important details discussed during the meeting can be entered here. --> Plan to have Meekail, Rachel, Delaney, Viktoria, Dr. Ojeme meet Thursday at 10 rather than Wednesday afternoon. Keep Friday at 11 as short status update. Discussed goals of Marcus's work and the "null" model group.