BANC connectivity analysis paper resources

Repository for data, code, figures, and manuscript assets for "Effective Connectome Alignment For Cell Typing and Identifying Sex Differences."

Repository structure

R/
  startup.R                                  — data loading from GCS + SeaTable
  panels-vnc-dimorphic-influence.R           — sensory → dimorphic influence heatmaps
  panels-vnc-morphology-connectivity-types.R — VNC type agreement & NBLAST panels
  banc-dimorphic-density.R                   — synapse density maps (VNC + brain)
  banc-vnc-type-changes.R                    — VNC type change CSVs + summary
  banc-vnc-type-review-images.R              — comparison PNGs for type-change review
settings/
  paper_colours_lacroix.csv                  — shared colour palette
data/
  banc/                                      — BANC-specific data & influence cache
  manc/, mcns/                               — cell inventory CSVs per dataset
  codex/                                     — drop-in inputs for type-changes script
  cache/                                     — GCS download cache (auto-created)
figs/
  figure_typing/                             — cell typing panels (.ai + linked PDFs)
  figure_dimorphic/                          — sexual dimorphism panels + density maps
    links/                                   — generated PDFs/PNGs

Analysis scripts

panels-vnc-dimorphic-influence.R

Computes sensory → dimorphic/sex-specific neuron influence for BANC (female) and maleCNS (male), producing paired heatmaps across three super_class groups (VNC intrinsic, ascending, descending). Influence is recalculated from scratch each run using the influencer package (PETSc solver). Outputs go to figs/figure_dimorphic/links/.

panels-vnc-morphology-connectivity-types.R

Compares VNC cell type assignments between morphology-based (NBLAST) and connectivity-based approaches, stratified by sexual dimorphism. Uses NBLAST scores from GCS and reviewed type-change CSVs from data/. Outputs go to figs/figure_typing/links/.

banc-dimorphic-density.R

Density plots of dimorphic / sex-specific synapses across BANC, MANC, FAFB, and maleCNS, registered to JRCVNC2018U (VNC) and JRC2018U (brain) reference spaces. Produces faceted density grids and BANC-vs-other subtraction maps for both pre- and post-synapses. Requires gsutil, nat.flybrains, nat.templatebrains, nat.jrcbrains, and the navis Python package (via reticulate) for maleCNS↔BANC coordinate transforms. Outputs go to figs/figure_dimorphic/links/. Adapted from bancpipeline/banc/update/banc-dimorphic-density.R; runs standalone.

banc-vnc-type-changes.R

Reads BANC seatable + reviewed CSVs, joins to franken_meta, identifies VNC intrinsic + effector type changes, computes hemilineage changes, looks up NBLAST scores per type-change category, and writes:

• data/vnc_type_changes.csv
• data/vnc_hemilineage_changes.csv
• data/vnc_effector_type_changes.csv
• data/mcns_effector_type_changes.csv

Optional input CSVs (BANC dimorphism, effector matches, LR mirror matches, alignment scores) live under data/codex/; sections that depend on them are skipped if files are missing. SeaTable update blocks from the original bancpipeline script have been intentionally omitted — re-add and uncomment banctable_update_rows() calls if you need to push back. Adapted from bancpipeline/banc/update/banc-vnc-type-changes.R; runs standalone.

banc-vnc-type-review-images.R

For each entry in data/vnc_type_changes_reveiwed.csv with accept_new = NA, generates a 3-panel PNG comparing the BANC neuron against the best MANC NBLAST match for both seatable_type and effective_type. Uses the BANC↔MANC NBLAST feather (GCS) for best-match selection with franken_meta fallback. Output PNGs go to figs/figure_dimorphic/links/. Adapted from bancpipeline/banc/update/banc-vnc-type-review-images.R; runs standalone.

Data sources

All connectome data is loaded from Google Cloud Storage at runtime:

• BANC metadata & edgelist: gs://lee-lab_brain-and-nerve-cord-fly-connectome/compiled_data/banc_888/
• maleCNS metadata & edgelist: gs://lee-lab_brain-and-nerve-cord-fly-connectome/compiled_data/malecns_09/
• NBLAST scores: gs://lee-lab_brain-and-nerve-cord-fly-connectome/nblast/
• SeaTable (BANC + maleCNS): queried via bancr::banctable_query() for columns not in the GCS feathers (e.g. sexually_dimorphic, malecns_cell_type)

Downloaded files are cached locally in data/cache/. Delete this directory to force a fresh download.

Dependencies

• R packages: bancr, influencer, malevnc, dplyr, arrow, reticulate, ComplexHeatmap, circlize, ggplot2, ggpubr, nat.flybrains, nat.templatebrains, nat.jrcbrains, nat.ggplot, Rvcg, rgl, fafbseg
• Python (via reticulate): petsc4py, gcsfs, pyarrow, navis + flybrains (for maleCNS↔BANC coordinate transforms)
• CLI: gsutil (for NBLAST + synapse feather/parquet downloads)

Running

# From the repository root:
Rscript R/panels-vnc-dimorphic-influence.R
Rscript R/panels-vnc-morphology-connectivity-types.R
Rscript R/banc-dimorphic-density.R
Rscript R/banc-vnc-type-changes.R
Rscript R/banc-vnc-type-review-images.R

Set BANC_VERSION to override the default data version (e.g. BANC_VERSION=banc_900).