Dataset DOI: https://doi.org/10.5061/dryad.v41ns1sc9

These files contain the data used to generate the figures in the associated iScience manuscript: two-photon calcium imaging of taste receptor cells (TRCs) and geniculate fibers, RNAscope in mouse and human taste tissue, and conditioned taste aversion tests. All data files are CSV. Cell and fiber numbers are specific to each file. A blank cell means no measurement was made, while 0 is a measured non-response. ΔF/F is a fraction, not a percentage.

Running the code

Place the data files in data/ and the scripts in scripts/ within one project folder, then run any script, for example python scripts/sweet_dose_curve.py. Scripts read from data/ and save figures to results/, which is created automatically. Requirements: Python 3.9+ with pandas, numpy, scipy, scikit-learn, and matplotlib (plus openpyxl for the preprocessing scripts in Section 2), and R with Seurat [version] and the packages loaded in UMAP_Code.R.

1. Concentration-response series

Each file has one row per cell and one column per concentration, with the unit in the header. Values are the peak ΔF/F during that stimulus window divided by the peak ΔF/F at the highest concentration tested in that cell. The highest concentration is therefore 1.00 for every cell; values above 1.00 mean a cell responded more strongly at an intermediate concentration. Each file is analyzed by the script of the same name (for example, sweet_dose_curve.py,cycloheximide_dose_curve.py, quinine_dose_curve.py, salt_dose_curve.py, sour_dose_curve.py, and umami_dose_curve.py).

Every umami concentration includes 1 mM IMP. Sour, salt, and cycloheximide were each run as two panels:

• Sour: cell1–7 at pH 6, 5.2, 4.6, 4, 3.5, 2.8, 2.2; cell8–17 at pH 4.8, 4.5, 4, 3.75, 3.5, 2.8, 2.2.
• Salt: cell1–10 at 10, 20, 30, 45, 60, 90, 120 mM (normalized to 120 mM); cell11–12 at 20, 60, 120, 360 mM (normalized to 360 mM).
• Cycloheximide: cell1–3 at 1, 4, 7, 10, 20, 50, 100 µM; cell4 at 0.1, 1, 10, 30, 50, 75, 100 µM.

Each script fits a four-parameter logistic, y = d + (a − d) / (1 + exp((c − x) / b)), to the mean response and reports c as the EC50 (in linear units) with R².

2. Taste panel imaging

V1_Taste_Panel.csv (39 TRCs), V2_taste_panel.csv (44 TRCs), and Fiber_taste_panel_data.csv (85 fibers) contain processed ΔF/F traces on a common 0.05 s grid from 0 to 275 s. Raw fluorescence traces are available upon request from the lead contact. Columns are time (seconds), Stimuli (the stimulus being presented, blank between presentations), and then Cell1_trial1, Cell1_trial2, and Cell1_average (mean of the two trials) for each cell, or Fiber1_… in the fiber file. Traces are blank after a recording ends.

Each stimulus lasts 5 s, with onsets at 30, 65, 100, 135, 170, 205, and 240 s. Presentation order:

• V1: Acek EC50, Acek 30 mM, cycloheximide, umami EC50, umami 30 mM, sour, salt
• V2 and fibers: Acek EC50, sour, umami EC50, cycloheximide, Acek 30 mM, salt, umami 30 mM

The three processing scripts document how these files were made. They need the raw data, so they cannot be run on the deposited files. df_f_for_trcs.py computes TRC ΔF/F with a separate F0 for each stimulus (mean of the 100 preceding frames), applied up to the next stimulus's baseline. df_f_for_fibers.py uses a single F0, the median of the seven baseline windows. interpolation_code.py Resamples each trace onto the 0.05 s grid by PCHIP interpolation. The columns were then renamed, the Stimuli column added, and the tables exported to CSV without changing any values.

3. RNAscope

Mouse files are CV_RNAscope.csv (3,093 cells), foliate_RNAscope.csv (1,550), and the fungiform files Tip_RNAscope_Exp1.csv (200) and Tip_RNAscope_Exp2.csv (250), which are pooled for the figure. Columns are Cell_index, T1R1, T1R2, and T1R3. The human file Human_rnascope_data.csv (20 cells) has columns Cell index, TAS1R1, and TAS1R2. Values are signal intensities normalized to 0–100, and a probe counts as expressed at 10 or above.

rnascope_figure.py keeps mouse cells that express any T1R, classifies them as sweet (T1R2+T1R3), umami (T1R1+T1R3), or savory (all three), and plots each class as a percentage of classified cells. human_rnascope.py classifies human cells as TAS1R1+/TAS1R2+, TAS1R1 only, TAS1R2 only, or neither.

4. Conditioned taste aversion

Sweet_CTA_Data.csv Covers mice conditioned against sweet (7 LiCl, 6 control) and tested for umami, salt, and sweet preference. Umami_CTA_Data.csv covers mice conditioned against umami (6 LiCl, 6 control) and tested for sweet, umami, and salt preference. Each 30-minute two-bottle test was run twice, with the tastant bottle on the right and then on the left, giving six blocks stacked vertically per file. Each block starts with a row giving the test, the bottle position, and the column labels. LiCl mice are listed first, then controls. Bottle weights and consumption are in grams, and the preference score is bottle 1 consumption divided by total consumption.

5. Single-cell RNA sequencing

Single-cell RNA sequencing data from mouse circumvallate taste tissue (10X Genomics, [16,903] cells) are in GEO under GSE345465. UMAP_Code.R Run the cluster analysis: place filtered_feature_bc_matrix.h5 an HDF5 file holding a single-cell RNA-seq count matrix (cells × genes, with only cell-containing barcodes kept) generated with Cell Ranger in data/ and run the script from the project folder.

6. In vivo imaging stage

Top_plate.stl and Bottom_plate.stl are design files for the stage that holds the tongue during in vivo two-photon imaging.