@rtscott2001 - great idea! Once the team settles on a meeting time, I’ll post the meeting invite on the calendar on the forum.
1 Like
Excellent, of not when you create a calendar invite
New Topic → AI/ML AWG Calendar
Then Gear Wheel → Create Event
Then date/title/url for mtg/ PRIVATE / and make sure to tag AIML AWG
1 Like
Interesting paper on this topic. @rtscott2001 @svcostes Can we create a topic that is not strictly focused on the AI/ML aspects of this project and have a place for related papers and discussion?
Interesting paper -
" DdrC, a unique DNA repair factor from D. radiodurans , senses and stabilizes DNA breaks through a novel lesion-recognition mechanism"
1 Like
Absolutely. Here is how to make a post as a general “discussion” in the generic AWG Discussion category
1 Like
Hi, do we have a new time and date set for the next meeting? Thanks!
Reminder that this subgroup is meeting in ~10 minutes. Message me privately if you are part of this subgroup and don’t have the Zoom info.
Agenda:
Since there are several efforts ongoing within this subgroup, we would like to invite everyone who has an ongoing effort to present their status at this next meeting. Please come prepared to address the following items:
-
What is the problem you are trying to solve? (radiation type classification, foci detection/segmentation, etc)
-
What is your approach to solving that problem?
-
What is your status? (model performance/accuracy, xAI status)
-
How close are you to being happy with your completed code and research product? What are your next steps?
We would like to start moving toward one or several manuscripts, so please keep that in mind while preparing for this next meeting.
3 Likes
I had a couple of questions before I start with developing a classifier (for X-ray dose).
- Are we (still) using only images with a maximum intensity of 4,000?
- Are we scaling the images, i.e. they all have values between 0-255, prior to feeding them into a classifier or deriving meta features from them?
- (for @walt ) Could you list the meta features you used once more. I downloaded the 2011 article you mentioned but wanted to make sure.
Thank you!
2 Likes
Hi @mboerrigter ,
- Yes
- Scale between vmin=400, vmax=4000
- Mean Phase Congruency (MPC)
- Entropy of Phase Congruency (EPC)
- Entropy of the Image (EDIS)
- Mean Gradient Magnitude of the Image (MGDIS)
2 Likes
Thank you Lauren!
1 Like
Is anybody aware of any other open source datasets featuring DNA damage and light microscopy for comparison purposes?
1 Like
Hi @mboerrigter I have been looking for one also and have not found one that is an exact fit for our use case. It has been a while since I looked though. I’ll try again over the holidays.
2 Likes
The DeepFoci paper here lists development of a dataset. They may be able to share this data upon request.
4 Likes
Hi Parham, We would be happy to have you join the team! send me your email address and I’ll add you to the list. We meet monthly on the second Wednesday of each month at 12 PM ET.
1 Like
Hi everyone,
I wanted to make sure the AI AWG Microscope subgroup is aware that the complete set of DNA damage imaging data in mice (OSD-366) is now available here:
NASA OSDR: Open Science for Life in Space
For each dose, time point, and animal, about 800 cells have been imaged. These are packaged into two zip files per condition—one for raw data and one for spot detection using the wavelet algorithm I developed.
- 15 strains
- ~5 animals/strain (3M, 3F, with some loss)
- 3 radiation types
- 3 dose levels (0, low, high)
- 4 time points (4, 8, 24, 48 hrs)
= ~2,700 conditions × 800 cells = ~2 million cells
The microscopy benchmark dataset we released on AWS a few years ago is just a subset of this. This new release includes full 3D stacks of 53BP1 in .ics and .ids formats (rather than projections), allowing for full 3D spot detection.
I’m excited to test the classifier we’ve been building on this expanded dataset, which now includes other radiation types like Argon. We should also be able to observe the Poisson distribution for Argon, though the dose increments are smaller—perhaps allowing us to distinguish it from Fe.
11 Likes
EPIC. Cheers and thanks Sylvain @svcostes 
I hope other AWG members @ALSDAawg @MultiOmicsAWG @AnimalAWG check out this gigantic imaging rodent spaceflight study 
3 Likes
Hi everyone,
Is there any meeting planned for that time? Thank you
Best Regards,
Fae Gaze
Biostatistician> Machine Learning Researcher| Computational Biologist
Advancing AI in biomedical research
Please note: Due to active research and collaborations, replies may occasionally be delayed.
3 Likes
Hi @Fgaze Fae,
Yes! This group on microscopy and DL has been meeting for a while.
I believe the next meeting is June 11th.
You should see the meeting invite in your calendar: https://awg.osdr.space/upcoming-events
You can ping-chat with @evartsb who leads the collaboration. I believe they are in the manuscript phase. Unsure
Cheers
-Ryan
Great news !! Big thanks @svcostes and GeneLab! Can’t wait to explore the new data! We can discuss this at our next meeting on 6/11!
@lauren.sanders @svcostes - What is the best way to programmatically access the data ? Via the OSDR API, or are there plans to add this data to AWS data registry?
1 Like
Interesting new paper providing insights into mechanisms of RAD51 HR:
eLife Assessment
“This landmark study describes the structure of the human RAD51 filament with a recombination intermediate called the displacement loop (D-loop). Using cryogenic structural, biochemical, and single-molecule analyses, the authors provide compelling evidence on how the RAD51 filament promotes strand exchange between single-stranded and double-stranded DNAs. The work will be of interest to the community of homologous recombination and DNA repair, as well as genome stability more generally.”
5 Likes
Hi, Thank you for leading this group
I have been working in this field. I am very interested in joining this group and would like to collaborate.