Experimental Quality Control Induces Changes in Allen Mouse Brain Connectomes

This repository contains the Supplementary Data from "Experimental Quality Control Induces Changes in Allen Mouse Brain Connectomes" (Nathan et al., 2026; in revision, Imaging Neuroscience). For code to reproduce all analyses from Nathan et al., 2026, see https://github.com/vik16nathan/allen_connectome_qc/).

We include the following files:

• Supplementary Data 1: Full QC images, including injection and projection images. Within the zipped file, there are folders knox_inj/ and knox_proj/ containing subdirectories of QC images at various binarization thresholds (knox_inj/bin0.01 and knox_inj/bin0.5 for injection images; knox_proj/bin0.01 and knox_proj/bin0.1 for projection images)
• Supplementary Data 2: QC ratings across all images and both raters. For more information on the rating scheme, see the Methods section of the manuscript and Supplementary Methods Figure 1 for representative examples. 
  • knox_inj_prod_bin0.5_qc_harmonized.csv, knox_proj_bin0.1_qc_harmonized.csv: Final, harmonized QC ratings for ALL experiments. 
  • vikram_steph_*_qc_differences.csv: Experiments where the two raters (V.N. and S.T.) initially disagreed on injection or projection QC ratings, as well as the initial ratings for both raters.
  • vikram_steph_*_qc_differences_harmonized.csv: Harmonized ratings for the experiments above after discussing disagreements.
• Connectomes: We attach all rebuilt connectomes after QC. The rows and columns correspond to region numbers (see knox_connectome_rgn_numbers_ccfv3.txt and oh_connectome_region_numbers_ccfv3.txt). These numbers use the CCFv3 labelling scheme from Wang et al. 2020 (see allen_ccfv3_tree_wang_2020_s2.xlsx, taken from the Supplementary Data of Wang et al., 2020). We use the following naming conventions:
  • The files with "knox" refer to the RVM, and the files with "oh" refer to the HM (see Nathan et al., 2026 and Knox et al., 2018)
  • The suffix "new" refers to after QC; "old" refers to before QC (the original data from Knox et al., 2018).
  • The connectomes with "ipsi" refer to regional connectivity ipsilateral to the injected region, whereas "contra" refers to connectivity contralateral to the injected region. Connectomes with "bilateral" concatenate ipsilateral and contralateral connectivity. 
  • Each connectome is calculated either for 291 gray-matter regions from Knox et al., or the 211 regions from Oh et al. (out of the original 213 regions) that are still included in the most recent release of the Allen Institute's CCFv3 atlas (Wang et al., 2020). Hence, 291 vs. 211 refer to different sets of regions from Knox et al. and Oh et al., respectively.  
    • We note that knox_conn_*_new.csv and knox_conn_*_old.csv use 291 regions, by default.
    • We also note that there are ~3 fewer contralateral regions than ipsilateral regions, since these regions have no connectivity to any of the injection sites. Hence, we pad the connectivity for these regions with zeros.
    • All connectomes use the output metric "normalized connection strength" (see Oh et al., 2014), except those that explicitly state "density," which refers to the normalized connection density (connection strength divided by the size of the source region).