• Date: 2026-05-26
• Time: 09:00AM (PT)
• Location: Teams Meeting

Agenda

• From Sarah Ruediger's group: Yuyan Huang - Standard-Oddball analysis
• From @rcpeene OpenScope databook use for Predictive processing.

Meeting Recording

Meeting Notes

Analysis of Synaptic Inputs in Mouse V1 During Visual Mismatch Paradigm: Yuyan presented an analysis of the OpenScope SLAP2 dataset, supervised by Sarah Ruediger Lee, focusing on how synaptic inputs in mouse V1 adapt during a standard mismatch paradigm, with active discussion and technical feedback from Stefan, Alexander, Jerome, and others.

Research Question and Dataset: Yuyan introduced the main research question: how synaptic inputs into V1 adapt during a visual mismatch paradigm, using data from five mice and five experiments, noting that two datasets lacked block 2 due to experimental design improvements.

Analysis Methods and Metrics: Yuyan described the calculation of response matrices, including absolute and relative (baseline-subtracted) responses, and explained the rationale for focusing on relative responses due to fluctuating baselines across trials.

Findings on Orientation Specificity: Yuyan compared ROI-wise responses to standard and oddball stimuli, finding that absolute response plots showed strong correlation between standard and oddball responses, while baseline-subtracted measures were less stable and did not reveal consistent orientation-specific effects.

Discussion of Regression Analysis: Stefan questioned the interpretation of regression slopes and the lack of forced zero intercepts, leading to a discussion about the limitations of linear regression in this context; Yuyan acknowledged the limitation and focused on slope differences, while Sarah Ruediger Lee and Jerome provided additional context and suggestions.

Baseline Analysis and Statistical Considerations: Yuyan presented baseline analysis, showing that baseline shifts were small and negatively correlated with mismatch response changes; Alexander and Stefan raised concerns about potential statistical artifacts due to the calculation method, leading to a consensus that while the finding is mathematically expected, it still provides insight into the relationship between baseline and evoked responses.

Contextual Effects and Heterogeneity: Yuyan explored context-dependent adaptation across blocks, finding that while some contextual changes affected response slopes, the results were heterogeneous and limited by sample size and methodological constraints.

Future Directions and Methodological Limitations: Yuyan concluded that responses at the dendritic level are heterogeneous and may be shaped by broader context rather than stimulus identity alone, suggesting future analysis could involve classifying ROIs into functional groups and addressing methodological limitations such as linear regression and anatomical specificity.

Technical Discussion on Baseline Correction and Statistical Artifacts: Alexander, Stefan, Sarah Ruediger Lee, and Jerome engaged in a detailed technical discussion with Yuyan about the implications of baseline subtraction in response analysis, highlighting potential statistical artifacts and suggesting alternative approaches for more robust analysis.

Concerns About Baseline Subtraction: Alexander pointed out that the negative correlation between baseline and response is a mathematical consequence of the calculation method, as the response is defined as the difference between absolute response and baseline, leading to a trivial relationship.

Alternative Analysis Suggestions: Participants suggested comparing absolute baseline and absolute evoked responses directly, and considering global versus local baseline fluctuations, to avoid confounding effects and better understand the biological relevance of baseline shifts.

Clarification of Baseline Calculation: Yuyan clarified that the baseline was computed as the average response during a 300 ms pre-stimulus window for each trial, and Sarah Ruediger Lee explained that this approach is standard in electrophysiology, though less common in calcium imaging.

Discussion of Noise and Internal State: Stefan and Alexander discussed the influence of noise and internal state fluctuations on baseline measurements, suggesting the use of bootstrapping or noise estimation to assess the robustness of baseline-dependent findings.

Development and Demonstration of OpenScope Data Analysis Notebooks: Carter presented two analysis notebooks for mesoscope and Neuropixels datasets in the OpenScope Data Book, demonstrating their structure, usage, and adaptability for various analyses, with input and questions from Farzaneh, Sarah Ruediger Lee, Alexander, Jerome, and Lucas.

Overview of OpenScope Data Book: Carter explained that the OpenScope Data Book is a public online library of analysis notebooks designed to work with NWB files from OpenScope projects, providing both introductory and advanced analysis tools.

Notebook Structure and Usage: Carter described the structure of the notebooks, which include sections for data access, metadata extraction, visualization, and basic analysis, and recommended running them locally with appropriate environment setup instructions.

Adaptability for Quality Control and Custom Analyses: In response to Farzaneh's question, Carter confirmed that the notebooks can be adapted for quality control or other specific analyses, and Jerome suggested that additional QC notebooks could be created and linked to data release papers.

Technical Details and Data Alignment: Carter demonstrated how to extract and align data from NWB files, including DFF traces, behavioral data, and stimulus epochs, and explained the use of SciPy interpolation to align independent time axes for different data streams.

Discussion on DFF Calculation and Data Packaging: Alexander and Jerome discussed the appropriate method for DFF calculation, noting differences in baseline fluctuations between cell types and the need for flexible approaches depending on the analysis; Carter noted that the final code for SLAP2 NWB packaging is still under development.

Neuropixels Notebook Features: Carter outlined the features of the Neuropixels notebook, including probe and electrode metadata, quality metrics, spike raster plots, behavioral alignment, LFP data visualization, and trial-based analysis of firing rates and z-scored responses.

Collaboration and Next Steps for Data Analysis Resources: Jerome proposed collaboration on developing analysis notebooks for SLAP2 and EFIS datasets, encouraged sharing of presentations and resources, and outlined plans for future meetings to discuss ongoing analysis and potential modeling approaches.

Notebook Collaboration and Resource Sharing: Jerome suggested that questions about baseline measurement and analysis should be addressed in the shared notebooks, and invited Yuyan and others to collaborate on developing these resources for the community.

Application to EFIS Data: Jerome noted that similar block structures in EFIS data allow for comparable analysis approaches, and encouraged Yuyan to apply her methods to this dataset.

Future Meeting Plans: Jerome announced that the next meeting agenda is not finalized, but may include updates from ongoing analyses or a presentation on modeling approaches related to EI balance in the data.