Data from: Local generation and efficient evaluation of numerous drug combinations in a single sample
Data files
Apr 30, 2026 version files 2.48 GB
-
02272023_uniform_CMTPX_DR.zip
142.09 MB
-
03082021.zip
446.31 MB
-
04072023.zip
51.16 MB
-
04212021_hyper7_time_course.zip
485.07 MB
-
05212021_t2.zip
578.82 MB
-
05212021_uniform_t1.zip
104.06 MB
-
060221021_controls_same_settings.zip
90.44 MB
-
060221021_point_5x.zip
87.91 MB
-
09262019_si655_tiles_masks.zip
345.85 MB
-
12292020_qds_8chamber_time_course.zip
32.49 MB
-
flow.zip
37.01 MB
-
Mathematica_PDF.zip
51.51 MB
-
Mathematica.zip
23.55 MB
-
README.md
10.35 KB
Abstract
We develop a method that allows one to test a large number of drug combinations in a single-cell culture sample. We rely on the randomness of drug uptake in individual cells as a tool to create and encode drug treatment regimens. A single sample containing thousands of cells is treated with a combination of fluorescently barcoded drugs. We create independent transient drug gradients across the cell culture sample to produce heterogeneous local drug combinations. After the incubation period, the ensuing phenotype and corresponding drug barcodes for each cell are recorded. We use these data for statistical prediction of the treatment response to the drugs in a macroscopic population of cells. To further the application of this technology, we developed a fluorescent barcoding method that does not require any chemical drug(s) modifications. We also developed segmentation-free image analysis capable of handling large optical fields containing thousands of cells in the sample, even in confluent growth condition. The technology necessary to execute our method is readily available in most biological laboratories, does not require robotic or microfluidic devices, and dramatically reduces resource needs and resulting costs of the traditional high-throughput studies.
Brief summary of dataset contents, contextualized in experimental procedures and results.
This dataset contains all raw data associated with the study, including fluorescence microscopy images (provided as ZIP archives), flow cytometry files, and Mathematica notebooks used for quantitative analysis and figure generation. Only small and moderate-sized imaging experiments are included to comply with Dryad upload limits. All microscopy images are raw, unprocessed files exported directly from acquisition software. Tile stitching (where applicable) was performed using Zeiss ZEN Blue. All downstream quantification and figure generation were performed using the Mathematica notebooks included in this dataset.
Data organization
All files are provided as ZIP archives. Each ZIP archive corresponds to a single experiment or data type:
flow.zip
Mathematica.zip
12292020_qds_8chamber_time_course.zip
04072023.zip
060221021_point_5x.zip
060221021_controls_same_settings.zip
05212021_uniform_t1.zip
02272023_uniform_CMTPX_DR.zip
09262019_si655_tiles_masks.zip
03082021.zip
04212021_hyper7_time_course.zip
05212021_t2.zip
Each ZIP archive contains one folder with raw image files in their original formats (.png, .jpg, .tif) or raw flow cytometry files (.fcs).
Microscopy image archives
Microscopy data are organized by experiment date and condition. Filenames follow one of two patterns:
Modern microscope format:\
Legacy slide‑scanner format:\
Channel definitions
Legacy slide‑scanner channels:
c1 = blue
c2 = green
c3 = red
c4 = far red
Modern microscope channels:
EGFP‑T1 = green fluorescence (EGFP)
LDFRe‑T1 = far red (Laser Dye Red)
T‑PMT‑T1 = brightfield
CeTCV‑T2 = blue
BY580‑T2 = red
AF405‑T3 = blue
QD655‑T3 = Qdot 655 probe
GFP‑405‑T2 = green (blue‑excited)
KiGRG = green
LTDR = far red
MarBl‑T3 = blue
cp = mask channel
Channels CTCRO‑T1 and CMDRP‑T2 were unused and are not interpreted.
Included imaging archives (ZIP files)
All imaging experiments are organized by acquisition date. Each folder contains:
- Raw microscope images (*.png)
- Derived segmentation masks (in binary/ subfolders, when applicable)
- Metadata embedded in filenames
- Channel identifiers (e.g., T-PMT-T2, LDFRe-T1, CTRed-T2, CTCRO-T1)
Filename structure
<experiment>s
Where:
- s# — sample index
- t# — timepoint
- m# — tile/field index
- channel — microscope channel (brightfield or fluorescence)
- ORG — raw, unprocessed export
12292020_qds_8chamber_time_course.zip
Raw images from an 8‑chamber time‑course experiment.
Representative filenames:
12292020_qds_8chamber_time_course_s1c1_ORG.png
12292020_qds_8chamber_time_course_s1c2_ORG.png
12292020_qds_8chamber_time_course_s1c3_ORG.png
11302020_qds_pre_transfect_GFP_t2_s7c4_ORG.jpg
04072023.zip
Live‑cell imaging (KiGRG, LTDR, T‑PMT).
Representative filenames:
04_07_2023_live_KiGRG_ORG.jpg
04_07_2023_live_LTDR_ORG.jpg
04_07_2023_live_T-PMT_ORG.jpg
060221021_point_5x.zip
5× magnification point scans.
Representative filenames:
060221021_point_5x_m01_CTRed-T1_ORG.png
060221021_point_5x_m01_EGFP-T2_ORG.png
060221021_point_5x_m01_CMDRP-T2_ORG.png
060221021_controls_same_settings.zip
Control samples imaged with identical settings.
Representative filenames:
060221021_controls_same_settings_s1m1_CTRed-T1_ORG.png
060221021_controls_same_settings_s1m1_EGFP-T2_ORG.png
060221021_controls_same_settings_s1m1_CMDRP-T2_ORG.png
05212021_uniform_t1.zip
Uniform illumination test, timepoint 1.
Representative filenames:
uniform_time_course_s1t01m02_CTRed-T2_ORG.png
uniform_time_course_s1t01m01_T-PMT-T2_ORG.png
uniform_time_course_s1t01m01_LDFRe-T1_ORG.png
02272023_uniform_CMTPX_DR.zip
CMTPX dye uniformity experiment.
Representative filenames:
02272023_uniform_CMTPX_DR_m01_LDFRe-T1_ORG.png
02272023_uniform_CMTPX_DR_m01_CTRed-T2_ORG.png
02272023_uniform_CMTPX_DR_m01_T-PMT-T2_ORG.png
09262019_si655_tiles_masks.zip
Mask files for si655 tile dataset.
Representative filenames:
09262019_si655_tiles_t001c123m01_ORG_cp_masks.png
09262019_si655_tiles_masks/
This folder contains segmentation‑related outputs derived from the raw si655 imaging dataset. These files were generated during the mask‑creation workflow and include:
*_mask.png— Binary segmentation masks (foreground = segmented object; background = non‑object). These correspond to the mask images already described in the README.*_outlines.txt— Text files containing the pixel coordinates of the segmentation boundaries for each object in the corresponding tile. Each line represents a contour, stored as a list of(x, y)coordinate pairs. These files are provided for users who wish to reconstruct object outlines programmatically or perform downstream shape analysis.*_output.png— Visualization overlays showing the segmentation mask applied to the original tile. These images are for quality‑control and illustration purposes only; they are not used in quantitative analysis. Typically, the underlying grayscale tile is shown with the segmentation boundaries or mask regions highlighted.
Together, these files document the full segmentation workflow for the si655 dataset, including raw masks, outline coordinate data, and visual QC overlays.
03082021.zip
Multi‑channel fluorescence imaging (CTRed, EGFP, AF405, QD655).
Representative filenames:
03082021_s1m001_CTRed-T1_ORG.png
03082021_s1m001_EGFP-T2_ORG.png
03082021_s1m001_T-PMT-T2_ORG.png
03082021_s1m001_AF405-T3_ORG.png
03082021_s1m001_QD655-T3_ORG.png
04212021_hyper7_time_course.zip
Hyper7 time‑course imaging.
Representative filenames:
04212021_hyper7_time_course_t01_EGFP-T1_ORG.jpg
04212021_hyper7_time_course_t01_T-PMT-T1_ORG.jpg
04212021_hyper7_time_course_t01_GFP-405-T2_ORG.jpg
05212021_t2.zip
Uniform illumination test, timepoint 2.
Representative filenames:
05212021_t2_m001_LDFRe-T1_ORG.png
05212021_t2_m001_CTRed-T2_ORG.png
05212021_t2_m001_T-PMT-T2_ORG.png
Flow cytometry data (flow.zip)
Contains raw .fcs files from a 96‑well plate experiment.
Well identifiers (e.g., A4, B1, D3) correspond to sample positions.
Example files:
04-Well-A4.fcs
04-Well-B1.fcs
04-Well-D3.fcs
Includes:
ExpSummaryForAPI.xml — instrument metadata summary
Mathematica notebooks (Mathematica.zip, Mathematica_PDF.zip)
Contains all Mathematica notebooks used for:
- fluorescence quantification
- flow cytometry analysis
- per‑cell metric computation
- manuscript figure generation
Because Mathematica is proprietary software, PDF versions of each notebook are provided in the Mathematica/Mathematica_PDF/ subfolder to ensure universal accessibility.
Notebook descriptions
- Figure_1c_1d.nb — Code and analysis steps used to generate Figure 1c and Figure 1d, including preprocessing, visualization, and statistical summaries.
- Figure_2.nb — Full workflow for generating Figure 2, including data loading, normalization, and multi‑panel plotting.
- Figure_3a_through_3d.nb — Analysis and visualization steps for Figure 3a–3d, including segmentation‑based quantification and multi‑condition comparisons.
- Figure_3e_3f.nb — Code for generating Figure 3e and 3f, focusing on downstream metrics and derived visualizations.
- Figure_4.nb — Processing and plotting steps for Figure 4, including summary statistics and graphical layout.
- Figure_6.nb — Analysis pipeline for Figure 6, including data filtering, curve generation, and composite figure assembly.
- Figure_7_8_S15.nb — Combined notebook generating Figures 7, 8, and Supplementary Figure S15, including multi‑dataset integration.
- Figure_S1_S2_S5.nb — Code for Supplementary Figures S1, S2, and S5, including preprocessing and visualization routines.
- Figure_S4_S5_S6_S7.nb — Notebook generating Supplementary Figures S4–S7, with associated analysis steps.
- Figure_S9_through_S14.nb — Code for Supplementary Figures S9–S14, including batch processing and multi‑panel figure assembly.
PDF versions of all notebooks are available in the Mathematica_PDF/ subfolder with matching filenames.
Tile stitching was performed using Zeiss ZEN Blue, not Mathematica.
Channel Description Update
CTCRO‑T1_ORG (04212021_hyper7_time_course)
The CTCRO‑T1 channel corresponds to the CellTracker Orange fluorescence channel used in the hyper‑7 time‑course experiment.
These images capture cytoplasmic staining used to track cell morphology and viability over time.
Binary Folder Descriptions
Several imaging folders contain a binary/ subfolder. These folders do not contain raw images.
Instead, they contain derived segmentation masks generated during analysis.
These masks follow the naming pattern:
<experiment>_m
The _cp_masks.png suffix indicates cell‑profile masks derived from the brightfield channel (T‑PMT‑T2).
No raw PNG files are duplicated in any binary/ folder.
3.1 05212021_uniform_t1/binary
Contains segmentation masks for each tile of the uniform time‑course experiment.
Masks correspond to the brightfield channel and are used for downstream quantification.
3.2 02272023_uniform_CMTPX_DR/binary
Contains segmentation masks for the CMTPX drug‑response experiment.
Masks correspond to brightfield images and were used to extract per‑cell fluorescence metrics.
3.3 05212021_t2/binary
Contains segmentation masks for the T2 replicate experiment.
As above, these are derived from the brightfield channel and used for cell‑level quantification.
Software requirements
- Zeiss ZEN Blue — tile stitching
- Mathematica 12+ — analysis and figure generation
- FlowJo (optional) — .fcs inspection
Reproducibility notes
- All images are raw exports from acquisition software.
- No compression, filtering, or downsampling was applied.
- All analysis notebooks are provided without modification.
- Representative filenames were generated automatically using a Python script.
Imaging data were obtained using an LSM 800 with an Airyscan confocal microscope (Zeiss).
Flow cytometry data were obtained using a CytoFLEX S flow cytometer (Beckman).
Submitted data contains imaging (PNG, JPG) and flow cytometry (FCS) files.
Data analysis was performed using proprietary Wolfram Mathematica software.
