The tuning_direction/ folder mostly consists of code to compute the direction/orientation tuning and spatial organization of mouse V1 presented in Figure 2, Supplemental Figure 1 and Supplemental Figure 2.
- estimate_selectivity_indexes.py estimate the direction and orientation selectivity indexes (DSI and OSI) of the recorded data as well as predictions from different models.
- extract_gabor_features.py extract directional Gabor grating parameters from the drifting Gabor stimuli in the Sensorium 2023 as their ground-truth was not shared.
- visualize_preference_differences.py visualize the difference in direction tuning preference between recorded and predicted responses (Figure 2D).
- visualize_selectivity_distribution.py visualize the OSI/DSI distributions.
- visualize_tuning_curve.py visualize the tuning curves of each neuron (Figure 2A).
- visualize_tuning_curve_correlation.py visualize the tuning curve correlations between recorded and predicted data (Figure 2B). visualize_tuning_width.py visualize tuning width (Figure 2C).
- estimate_tuning_similarity.py estimate the tuning similarity of the neurons as described in Ringach et al. 2016. The results are saved to
parquetfiles for quicker plotting. - visualize_tuning_similarity.py visualize tuning similarity, as computed in estimate_tuning_similarity.py, as Ringach et al. 2016 (Supplemental Figure 1).
- visualize_microcolumns.ipynb visualize micro-columns in a 3D scatter plot based on their direction tuning preference. You can filter neurons by OSI/DSI, xyz coordinates, etc.
- visualize_tuning_similarity_by_plane.py visualize tuning similarity by neuron plane (depth or z-axis).
- The model weights trained on the Sensorium 2023 challenge are available at huggingface.co/bryanlimy/ViV1T-closed-loop.