This dataset contains raw and processed image files used in Kang et al 2025 (https://doi.org/10.1101/2025.02.19.639023)

There are total of 16 CODEX samples at various time and media conditions. In the manuscript, samples with the standard media condition were published (Coverslip1, 4, 7, 13, 16). Full list of sample name is shown below:
    '101222_D10_Coverslip1',
    '102022_D10_Coverslip2_Reimage',
    '101422_D10_Coverslip3',
    '102222_D10_Coverslip4_Reimage',
    '101922_D10_Coverslip5',
    '102422_D21_Coverslip7',
    '102522_D21_Coverslip8',
    '102622_D21_Coverslip9',
    '102822_D21_Coverslip11',
    '102922_D21_Coverslip12',
    '103122_D28_Coverslip13',
    '110122_D28_Coverslip14',
    '110222_D28_Coverslip15',
    '110622_D28_Coverslip16',
    '110322_D28_Coverslip17',
    '110722_D28_Coverslip18'

File Hierarchy is described below:
raw
    - '101222_D10_Coverslip1'
        - Cycle_1_Field_1
        - Cycle_1_Field_2
              .
              .
              .
    - '110722_D28_Coverslip18'

processed
    - '101222_D10_Coverslip1_Processed'
        - tif
        - reg_Cyc_N
        - reg_bin_Cyc_N
        - tmat_Cyc_N
        - merged
        - max
        - max_clean
        - mp_score
        - mp_score_max
        - Cellpose_Segmentation_masks
        - intermediate_outputs
        - cp_output_cellpose
        - adata_Cellpose
              .
              .
              .
    - '110722_D28_Coverslip18_Processed'

metadata
    - markers.csv
    - clean_markers.csv
    - PROCODE_gRNA.csv
    - codebook.csv

analysis
    - adata_integrated
        - D10_STD_adata.h5ad
        - adata_Cellpose_Subset_All_Channels_011425.h5ad
    - Fig4_Results
        - ks_result_store
        - pairwise_clf_results_auc.pkl
    - arborization_outputs
        - morphology_data_compiled.csv
        - ttest_df.csv


#################################### 
"raw" folder contains all raw image in .ims format (imaris) generated by Andor Dragonfly Fusion software. Under each sample (coverslip), "Cycle_N_Field_M" folder contains 25 raw .ims files (4-channel Zstacked images), total of 225 image files at N-th cycle and M-th field. 
####################################
"processed" folder contains all processed images and associated intermediate outputs.
"tif" : converted .tif files from raw .ims file
"reg_Cyc_N" : registered image, using transformation matrix stored under "tmat_Cyc_N" calculated with binarized images stored under "reg_bin_Cyc_N"
"merged": Stacked, Z-stacked images from "reg_Cyc_N" where all cycles are concatenated. Edges are cropped to the smallest size among the cycles due to differences in transformation at each cycle. (10 z stacks, 38 channels, Y, X) and YX are typically ~2000 pixels. See "metdata/markers.csv" for cycle/channel-level information.
"max": maximum intensity projected image, from "merged", but duplicate DAPI channels (except the first cycle) and blank channels were removed from "merged" (25 channels, Y, X). 
"max_clean": Background subtracted, quantile-normalized images of "max", to be used as inputs to Matching Pursuit Deconvolution
"mp_score": Output of Matching Pursuit Deconvolution, deconvolving 7-channel epitope images into 22-channel ProCode images. Channel-level metadata is available in "metadata/codebook.csv" (channels 1-22 corresponds to columns left-right). "metadata/PROCODE_gRNA.csv" dictates which ProCode (i.e. AA5) corresponds to which gRNA (i.e. MAP4K4)
"mp_score_max": maximum-intensity projected image, from "mp_score"
"Cellpose_Segmentation_masks": Soma and nuclei segmentation mask, formatted to use as inputs to Cell Profiler Pipeline. Raw segmentation masks are stored under "Cellpose_Segmentation"
"mask": (optional) folder that masks out microscope-read/write error ROIs. 
"intermediate_outputs": Single-soma level output of barcoding pipeline. "CoversipN_soma_df_CellPose.csv" contains features related to epitope tag measurements, to be used as inputs of barcoding classifier. "CoverslipN_soma_final_Cellpose.csv" contains the result of barcode-assignment pipeline. 
"cp_output_cellpose": Single-soma level output of Cell Profiler Feature extraction pipeline. "CP_Cellpose_Soma_filtered_32.csv"
"adata_Cellpose": Combined data file that contains cell-level metadata and morphological feature output of Cell Profiler, at various processing levels (see github codes for detail)
"morphology_data": intermediate pickle files that extract neuronal arborization features from mp_score_max images. 
    Neurite_dfs.pkl : Python dictionary storing Dataframe of FOV by Genotype (ProCode ID), where each entry is the sum of detected neurite pixels. Each key refers to different filtering threshold.
    Branch_dfs.pkl : Python dictionary storing Dataframe of FOV by Genotype (ProCode ID), where each entry is the sum of detected number of branches. Each key refers to different filtering threshold.
    morphology_data.pkl : Dataframe same as "CoverslipN_soma_final_Cellpose.csv" with number of trunks for each single-cell is appended. 
    data_dfs.pkl : Field-level data aggregating FOV-level data (rep_data_dfs), normalized with the number of cells in each Field. 
####################################
"analysis" folder contains combined analysis output file aggregating sample-level data. 
"adata_integrated/adata_Cellpose_Subset_All_Channels_011425.h5ad": Compiled data across all samples. This is our extended dataset that is not included in the manuscript. 
"adata_integrated/D10_STD_adata.h5ad': Compiled data used in the manuscript - single-soma morphological fingerprint data (CODEX multiplexed channel with Cell Profiler features) at Day 10 timepoint, with standard media condition. 
"Fig4_Results": contains data analysis output files "ks_result_store" and "pairwise_clf_results_auc.pkl" 
"ks_result_store" is a nested dictionary where each key refers to each condition (Time_Media). Within each condition, non-target vs gene KS-test result dataframe is stored (rows are feature names, columns are p-value and associated statistics). "pairwise_clf_results_auc.pkl" is a dictionary where each key refers to each condition. For each condition, gene-by-gene matrix is stored, where each element is AUC of ROC curve of pairwise classification.
"arborizaiton_outputs": Field-level (technical replicates) "morphology_data_compiled.csv" compiles arborization features and sample-level data into a single, long-form dataframe. 
"ttest_df.csv": t-test results comparing non-target vs each gene, for each feature and time/media condition separately.