Data from: Spatial biases in visual feature representation of mouse dorsal lateral geniculate nucleus boutons
Data files
Jul 13, 2026 version files 3.12 GB
-
figure_data.zip
3.12 GB
-
README.md
3.77 KB
Abstract
Humans and other mammals show spatial biases in their perception of visual features, with the primary visual cortex (V1) considered a key source. However, earlier structures may also contribute. Recent work shows feature biases in the retina, but little is known about how they evolve as signals flow through the dorsal lateral geniculate nucleus of the thalamus (dLGN). Using in vivo calcium imaging, we investigated spatial frequency, orientation, direction, and temporal frequency representations in both the retina and in dLGN boutons. We found modest location-dependent biases in the average representation of each feature across visual space for boutons, while such biases were weaker in the retina. dLGN representations emerged from functionally and anatomically defined bouton subsets. Selective ablation of cortical feedback to dLGN modulated feature biases but did not eliminate them. Together, these results suggest that dLGN integrates retinal and cortical inputs to create spatially biased feature maps for V1.
Dataset DOI: 10.5061/dryad.tmpg4f5f9
Description of the data and file structure
The dataset is to generate panels for main figures in Cha et al.'s 'Spatial biases in visual feature representation of mouse dorsal lateral geniculate nucleus boutons' (2026).
figure_data.zip contains three folders, each of which is for widefield imaging data ('data_figure_epi', Fig 1 and 5), 2P imaging data ('data_figure_2p', Fig 2 and 3), and ex-plant retina imaging (2P) data ('data_figure_rgc', Fig 4). All data are saved in MATLAB data file format (.mat).
A. data_figure_epi
A1. Files named centroidParameters_[feature domain name].mat
These files contain the data for Figures 1 and 5 in the manuscript.
A1.1. Struct data array "FigureDataArray"
- The array contains any data unit for a mouse of a genetic line under a condition, and one can select a particular element by choosing the 'mouseId', 'groupName' and 'dataGroupName'.
- Field 'groupName' is either 'Wt1+dlgAxG8s' for wild type or 'Nt1+dlgAxG8s' for Ntsr1-cre line.
- Field 'dataGroupName' carries either value of 'Anes' for control or 'Anes+Dtx' for post diphtheria toxin injection. (Note that 'Awak' was not used in the study but a pilot dataset).
- Field 'parameterName' corresponds the feature domain name
- Field 'parameterMap' contains the preferred feature value, which corresponds to the 'feature map' in the manuscript (spatial frequency in cyc/deg, temporal frequency in hz, and orientation/direction in radian; visual angles/distances are in degree)
- Field 'FeatureGradientModel' contains the linear fitting data of feature deviation against visual field distance
- Visual field positions and distances are all in visual angle in degree.
A1.2. Struct array 'ReferenceMapDataArray'
- This array provides the alignment of V1 and retinotopic maps (azimuth and elevation) for each of 36 animals.
A2. Files named selectivityScores2_[feature domain name].mat
These files are structured the same as the file above, but they are for 'selectivity' analysis.
B. data_figure_2p
FieldDataArray.mat file contains a struct array FieldDataArray. The elements of the array correspond to the 67 2-photon imaging fields.
B1. Fields named 'depth' and 'fieldTag'
These respectively indicate the distance from the pia in um and the columnal position index of a given imaging session.
B2. Fields 's2pMeanImage' and 's2pRoiImagePositions'
They respectively contain the average field image and ROI positions computed by Suite2p.
B3. Field 'TuningDataArray'
This array contains the tuning analysis data. Each of the 6 elements corresponds the stimulus/feature domain whose tuning is computed. The subfield 'centroidParameterValues' contains the best feature value used in the manuscript for each ROI. Units are indicated in A1.1 above.
C. data_figure_rgc
RetinaData.mat contains a struct array DataTuningAll, each element of which contains data for each of 9 retinas. Fields SGSF, SG, DG, and CH contain 'best feature values', corresponding to 'centroidParameterValues' in B3 (see above) for spatial frequency, orientation, direction, and temporal frequency, respectively. Units are the same as in A1.1 and B3 above.
The field 'age' has values in days. The field 'eye' indicates the side which is either left ("L") or right ("R"). The field 'agngleCorrectionVUp' indicates the angle in degree to be rotated to make the ventral side up.
Code/software
The code for figure generation is shared in the GitHub link (See Related Software).
