Meeting Notes
- Date: 2025-11-25
- Time: 09:00AM (PT)
- Location: Teams Meeting
- Presentations: Presentation on newer datasets collection on Neuropixels and SLAP2 platform
Agenda
- Reminder of plans for data collection. @jeromelecoq
- Introduction to new pilot datasets collected on Neuropixels platform @severine2305
- Update on data collection ongoing on SLAP2 platform by @jeromelecoq
- Discuss plan for data analysis of pilot data.
Meeting Recording
Meeting Notes
Overview of Data Collection Plans and Experimental Design: Jerome provided an overview of the community project's data collection plans at the Allen Institute, outlining the experimental design involving multiple session types and cohorts, and highlighted the timeline and goals for pilot and production data collection.
Experimental Session Types: Jerome described four types of experimental sessions: standard mismatch, sensory motor mismatch, sequence mismatch, and duration mismatch, each designed to probe different prediction contexts in mice using drifting gratings and varying session parameters.
Cohort and Platform Structure: The plan involves five cohorts: two for neuropixels, two for two-photon imaging, and one for the SLAP 2 platform, with each cohort trained in either sensory motor or sequential contexts, and the SLAP 2 platform focusing on sensory motor exclusively.
Timeline and Data Collection Goals: Initial pilot data collection was scheduled for October and November, with the aim to inform any necessary modifications before full production in January, and the team is working to maintain this timeline.
Neuropixel Pilot Data Collection and Technical Challenges: Severine reported on the neuropixel pilot data collection, detailing the experimental procedures, probe targeting, data quality, and technical challenges such as probe breakage and opto-tagging issues, with contributions from Ryan Gillis and Casey Lennon-Jones.
Pilot Experiment Execution: Two pilot experiments were conducted using SST AI 32 mice, each involving four days of recording with six probes (except for two sessions with one missing probe), and included habituation protocols and post-experiment processing steps such as perfusion and tissue clearing.
Probe Targeting and Area Identification: Severine explained the targeting strategy for different brain areas (e.g., ACA, VSP, LM, MOP) using ISI maps and implant schematics, noting successful targeting in some cases and the need for further anatomical confirmation in others.
Data Quality and Stability: Six out of eight sessions passed quality control, with stable firing rates and low probe drift (below 50 microns), and the data from these sessions have been uploaded and made public.
Opto-Tagging and Genotype Investigation: Opto-tagging did not yield expected responses in the pilot mice, prompting tests with different fibers, increased power, and genotype verification; further recordings with new mice from a different litter are planned to resolve the issue.
Discussion on Wheel Use in Mouse Cages: Severine, Jerome, Karim, Jordan, and others discussed whether to place running wheels in mouse cages prior to experiments to improve running behavior during sensory motor sessions, considering concerns about context-specific learning and standardization.
Rationale for Wheel Introduction: Severine and others observed that pilot mice did not run much during experiments and proposed adding wheels to cages to habituate mice to running, potentially improving engagement during sensory motor tasks.
Concerns and Counterpoints: Sarah expressed concern that pre-exposure to wheels might affect visual-motor mismatch learning, while Jordan argued that mice can distinguish contexts and that wheel use in the cage is unlikely to interfere with experimental learning.
Technical Suggestions and Feasibility: Karim suggested adding friction to cage wheels to limit speed, but Severine noted the practical difficulty of implementing such modifications at scale; the group agreed to continue the discussion on the forum and seek consensus.
Data Access, Packaging, and Metadata Structure: Carter presented the procedures for accessing, processing, and analyzing the neuropixel and SLAP 2 data, including the use of NWB files, DANDI, Code Ocean, and the structure and content of metadata and experimental notes.
Data Availability and Formats: Six neuropixel sessions are available on DANDI in NWB format, with detailed experimental notes and a Jupyter notebook provided to guide users in accessing and analyzing the data.
Key Data Components: The NWB files include subject metadata, stimulus tables, opto-tagging data, receptive field mapping, control blocks, trial types, LFP, unit tables with spike times and quality metrics, eye tracking, and running speed.
Raw Data and Code Ocean Pipeline: Raw and partially processed data, including behavior videos and Kilosort outputs, are accessible via a public Code Ocean collection, and future plans include making the data processing pipeline more transparent and shareable.
Metadata and Probe Area Mapping: Metadata JSON files on Code Ocean contain probe-to-area mappings, but this information is not yet included in the NWB files; the team agreed to provide a summary table on the forum to clarify probe targeting for analysis.
SLAP 2 Data Collection Progress and Plans: Jerome updated the group on the SLAP 2 platform's data collection, reporting on the status of glutamate and voltage imaging experiments, encountered technical delays, and outlined the schedule for completing additional recordings.
Current Status and Delays: Data collection for SLAP 2 is ongoing, with two months of sensory motor context data collected; technical issues with the microscope and health monitoring of mice caused some delays, and not all sessions could maintain recordings from the same neurons.
Imaging Modalities and Future Plans: The project involves separate cohorts for glutamate and voltage imaging, with no current plans for dual imaging in the same cells; additional recordings are scheduled after the holiday break, and data packaging is a forthcoming priority.
Analysis Priorities and Coordination: Jerome, Carter, Severine, and others discussed analysis priorities for the neuropixel and SLAP 2 data, including temporal alignment, receptive field mapping, and context-specific responses, and agreed to coordinate analysis efforts via the forum to avoid duplication.
Key Analysis Questions: The group identified several analysis priorities: verifying temporal alignment, comparing receptive field mapping stimuli (Zebra, Trippy, Gabor), evaluating context-specific responses, and examining peripheral cortex activity.
Coordination and Transparency: To maximize productivity and avoid redundant work, participants agreed to use the forum to organize analysis interests, share progress, and form collaborative plans, with the goal of providing comprehensive feedback on the pilot data.
Probe Area Identification for Analysis: Severine and Carter clarified that probe-to-area mapping is available in metadata JSON files, and a summary table will be posted on the forum to assist analysts in selecting the correct probes for specific brain areas.