v1.12.0: protein-level test in an independent spinal-cord proteome (Amendment 17, W75)

This is the release cited by the submitted manuscript.

New: W75, protein-level test (Amendment 17)

scripts/run_w75_spinal_cord_proteome.R and docs/analyses/W75_spinal_cord_proteome.md test the ALS-up module proteins in an independent human spinal-cord proteome.

• Data: PRIDE PXD067060 (Feneberg et al. 2025; Oxford Brain Bank), detergent-soluble fraction. 14 ALS, 7 control, 6 PD and 9 AD donors.
• Specification: the analysis was specified and committed before any expression value was read.
• ALS vs control:
  • ANXA1: log2FC 1.84, FDR = 0.002.
  • C3: log2FC 1.63, FDR = 0.043.
  • GPNMB: log2FC 1.00, P = 0.39. Higher in direction, but quantified in only 14 of 21 samples.
• Module set test: could not be computed, because only 1 of the 3 quantified proteins was measured in every sample (deviation recorded in S3).
• Other diseases: GPNMB separated ALS from PD (P = 0.002). ANXA1 was also raised in PD and AD.

Supplementary Tables

• S1: Amendment 17 added.
• S2: 15 W75 rows added.
• S3: two W75 deviations added.
• S4t (new): module-protein results.

README

The README adds the data source, the source-data layout and pipeline step 12i.

v1.11.0: journal-style figures and tables (condensed Fig. 5, Fig. S1 audits, graphical abstract)

Release cited by the manuscript revised to Molecular Neurodegeneration author guidelines and house style. No analysis results changed; this release covers figures, tables and their wording.

Figures

• Fig. 5 (single-nucleus design effects) now has four panels for the ALS-up module only: Pineda processing date, Ulm age, Li depth, and the ALS-up forest plot (fig4_meta.R adds pa_up).
• Supplementary Fig. S1 is now "cohort-design audits". It keeps the discovery audit (a-d) and gains the panels moved out of Fig. 5: processing-date variance (e), the cross-cohort replication matrix (f) and the two-module forest plot (g).
• Fig. 4d has an Interpretation box.
• Graphical abstract. compose_figures_v3.R writes results/Figures_v3/Graphical_abstract/graphical_abstract.png (920 × 300 px).
• Labels. Panel labels are reworded throughout: "ALS-up module", "consensus up genes", "fixed modules", and "Analysis plan" in the Fig. 1 legend key.

Supplementary Tables

• In reader-facing text in S1-S3 and the README sheet, "Amendment N" now reads "analysis document N". Repository file paths keep their names, so the SHA-256 values in S1 still match docs/analysis_plans/.
• S1 is renamed Supplementary_Table_S1_analysis_plan_documents.tsv.

v1.10.0: within-lineage source test (Amendment 16) and single-module figure set

Release cited by the simplified manuscript, which follows one signature (the frozen ALS-up module): bulk replication, ventral-horn localisation, a myeloid-led multicellular state, and the single-nucleus design lesson.

New

• Amendment 16 and W74 (scripts/run_w74_within_lineage_source.R, docs/analyses/W74_within_lineage_source.md): a within-lineage test of whether the ALS-up signal is shared with the microglial compartment. The discovery microglial compartment showed a small same-direction shift, mainly GPNMB (fry P = 0.021, median log2FC 0.17, against 2.33 in astrocytes). The astrocyte-minus-microglia interaction was larger in astrocytes in Zelic and not distinguishable in Ulm and Pineda BA9 (combined signed z = 1.84, P = 0.066). The prespecified overall reading is mixed, and the result is not used in the manuscript conclusions. It is listed in Supplementary Tables S1 and S2.

Changed

• Figures: Fig. 2 now shows the frozen-module tests (a) and gene-level fold changes (b). The consensus-signature panels moved to Supplementary Fig. S3a,b, and the clinical association to S3c. Fig. 3a shows the ALS-up module only. Fig. 1a no longer lists the consensus signature.
• compose_figures_v3.R: fixed source-data directory names (Fig. 2 and Fig. 5 source data were previously not copied). Each figure now receives only the source data of its own panels.
• Supplementary Tables S1 (Amendment 16) and S2 (one W74 row).

v1.9.0: generic-signature comparison (Amendment 15) and revised figure order

Adds Amendment 15 and the figure order of the revised manuscript, "Cross-platform validation of an ALS glial transcriptomic signature reveals a ventral-horn myeloid–glial response".

New in v1.9.0

• Amendment 15 (docs/analysis_plans/..._AMENDMENT_15.md), committed before W73 was run. It asks whether the ventral-horn response is generic neuroinflammation.
• scripts/run_w73_generic_signature_comparison.R (W73). It tests six msigdbr sets, with module genes removed, in the two territory tests: inflammatory response, TNFα–NF-κB, interferon-γ, complement, a TYROBP microglial network and an ageing-brain set. It then refits the ALS-up module with each set's score as a covariate.
  • All six programmes rise more in the ventral than the dorsal horn in ALS (combined one-sided P from 3.7 × 10⁻⁵ to 0.004).
  • The module is distinct from inflammatory, TNFα and interferon-γ signalling.
  • It partly overlaps complement and ageing programmes.
  • It is largely shared with the TYROBP network.
  • Disease specificity was not tested.
• Supplementary Tables:
  • S1: Amendment 15.
  • S2: W73 tests.
  • New sheets S4r and S4s.
• Figures.
  • Main figures renumbered to the revised Results order: Fig. 2 bulk replication, Fig. 3 ventral horn, Fig. 4 cell source with summary schematic, Fig. 5 single-nucleus reproducibility.
  • The Fig. 1 commit-timeline panel was removed.
• README and CITATION (v1.9.0) updated.

No data are included.

v1.8.2: figure order and title for the revised manuscript

Figure order and article title updated to match the revised manuscript, "Cross-cohort validation of an ALS glial transcriptomic signature identifies a ventral-horn myeloid–glial response".

Changes in v1.8.2

• scripts/figures_v2/compose_figures_v3.R

  • Fig. 4 is now the ventral-horn response (two spatial cohorts).

  • Fig. 5 is now the cell-source figure, with a new summary schematic (panel d).

  • Supplementary figures are renumbered in order of first citation:

    Old	New
    S3	S8
    S4	S3
    S5	S4
    S6	S5
    S7	S6
    S8	S10
    S10	S7

    S1, S2 and S9 are unchanged.

• README: the article title is updated.
• CITATION: updated to v1.8.2.

Analysis scripts, results and Supplementary Tables are unchanged from v1.8.1. No data are included.

v1.8.1: updated Fig. 5c working model

Figure update for the revised manuscript.

Changes in v1.8.1

• scripts/figures_v2/compose_figures_v3.R: the Fig. 5c working model now shows three results:
  • the activated-myeloid co-variation of the module (W70);
  • the disease-state expression of the leading genes in ALS spinal cord (GPNMB and C3 microglial, LTF astrocytic, ANXA1 and EMP3 vascular; W72);
  • the protein pattern (CD68 and HLA-DR, not CD45 or CD11b; W68).
• CITATION updated to v1.8.1.

Analysis scripts, results and Supplementary Tables are unchanged from v1.8.0. No data are included.

v1.8.0: SOD1-G93A time course and disease-state cell source (Amendment 14)

Adds Amendment 14: a spatial time course in SOD1-G93A mice, a third human spatial cohort, and the disease-state cell source of the module genes in human spinal cord.

New in v1.8.0

• Amendment 14 (docs/analysis_plans/..._AMENDMENT_14.md), committed before W71 and W72 were run.
• W71: Maniatis et al. 2019 ALS-ST data
  • Scripts: build_w71_maniatis_pseudobulk.py, run_w71_mouse_time_course.R, run_w71_human_colocalisation.py.
  • Mouse, SOD1-G93A vs SOD1-WT lumbar cord, ventral vs dorsal horn at P30, P70, P100 and P120. Only 2-4 module orthologs are testable, so the mouse module is mainly C3.
    • Cross-species replication reading: nominal (P100 interaction P = 0.043, BH = 0.17).
    • Timing reading: symptomatic or late. There is no change at P30 or P70; the rise starts in the ventral horn at P100 and extends to the dorsal horn at P120.
  • Human: seven ALS patients, no controls, so analysis is descriptive. The module score correlates most with microglial scores, but weakly.
• W72: disease-state lineage share in GSE287257
  • Scripts: run_w72_zelic_disease_cell_source.py, run_w72_lineage_de.R.
  • GPNMB (74%) and C3 (81%) come mainly from microglia in ALS spinal cord, LTF from astrocytes, and ANXA1 and EMP3 from vascular cells.
  • ALS also raises GPNMB and C3 in astrocytes.
• scripts/figures_v2/figS10_mouse_cell_source.R: new Supplementary Fig. S10. The Fig. 1 timeline is updated.
• Supplementary Tables:
  • S1: Amendment 14.
  • S2: W71 tests, 482 rows in total.
  • S3: one implementation detail.
  • S4: new sheets S4n-S4q.
• README (data source and pipeline) and CITATION (v1.8.0) updated.

The Maniatis data come from the ALS-ST portal (https://als-st.nygenome.org); see the README for the expected folder layout. No data are included.

v1.7.0: microglia-adjusted ventral-horn test (Amendment 13)

Adds Amendment 13: the territory × disease test of the ventral-horn response, adjusted for microglial content.

New in v1.7.0

• Amendment 13 (docs/analysis_plans/..._AMENDMENT_13.md), committed before W70 was run. It asks whether the ventral-horn rise of the frozen ALS-up module is explained by microglial abundance.
• scripts/run_w70_microglia_adjusted_territory.R (W70): refits the lumbar (GSE330224, W67 inputs) and cervical (GSE288365, W68 inputs) territory models with a centred unit-level microglial marker score.
  • Primary score: all microglial markers. Sensitivity scores: homeostatic-only and activated-only markers.
  • Reports the attenuation against the unadjusted interaction, the prespecified Stouffer combination and the reading rule.
• Result (docs/analyses/W70_microglia_adjusted_territory.md):
  • Frozen ALS-up module: attenuation 53% (lumbar) and 75% (cervical); combined one-sided P = 0.020; reading "partly shared with microglial abundance".
  • Most of the shared part is carried by the activated-myeloid markers.
  • Consensus up arm: not explained by microglial content.
• scripts/figures_v2/figS9_microglia_adjusted.R: new Supplementary Fig. S9, added to compose_figures_v3.R.
  • Fig. 1 timeline and evidence map updated.
• Supplementary Tables: S1 (Amendment 13), S2 (W70 tests), S4l and S4m added.
• README and CITATION (v1.7.0) updated.

No data are included.

v1.6.0: streamlined figure set for the revised manuscript

Streamlined figure set for the revised manuscript, "Prespecified cross-cohort testing localises an astrocyte-compartment ALS signature to a multicellular myeloid–glial response in the spinal ventral horn".

New in v1.6.0

• scripts/figures_v2/compose_figures_v3.R: assembles the submitted figure set from saved panel objects.
  • Five main figures: design; single-nucleus replication and confounders; bulk hold-out; cell source; ventral-horn response with a working model.
  • Eight supplementary figures.
  • The source data of each contributing panel are copied next to the assembled figure.
• Figure scripts: each now saves its panel objects (save_panels in fig_common.R). Two panels store their annotation positions as data columns, so that saved panels are self-contained.
• README (figure workflow and article title) and CITATION (v1.6.0) updated.

Run the figures_v2 scripts first, then compose_figures_v3.R. Analysis scripts are unchanged from v1.5.0. No data are included.

v1.5.0: independent spatial replication (W68) and dose-response (W69)

Adds the independent spatial replication (Amendment 11), the dose-response and cross-cohort analyses (Amendment 12), Fig. 7 and Supplementary Fig. S6 of the manuscript, "Cross-cohort donor-level analysis reveals a reproducible multicellular glial–vascular ALS transcriptional programme detected in astrocyte compartments".

New in v1.5.0

• docs/analysis_plans/..._AMENDMENT_11.md: replication of the motor- vs sensory-territory test in GSE288365 (Visium CytAssist cervical spinal cord with a protein panel). Fixed before any expression summary was computed.
• scripts/extract_w68_gse288365_metadata.R, build_w68_gse288365_pseudobulk.R, run_w68_replication.R, run_w68_protein_colocalisation.R: author spot annotations, donor x territory pseudobulk, territory x disease interaction (donor-blocked limma-voom; fry, camera), motor-neuron proximity, and protein co-localisation.
• docs/analysis_plans/..._AMENDMENT_12.md, scripts/run_w69_dose_response_and_combination.R: within-ALS dose-response of the motor-territory response with a motor-neuron index (with an expression-matched random-set control), and Stouffer combination of the two territory tests (post hoc summary).
• scripts/figures_v2/fig7_motor_territory.R, figS6_dose_response.R: Fig. 7 and Supplementary Fig. S6. Figs. 1 and 6 updated.
• docs/analyses/W68_*.md, docs/analyses/W69_*.md; Supplementary Tables S1-S4 updated (S4i-S4k added); README and CITATION (v1.5.0) updated.

All other scripts from v1.4.0 are unchanged. No data are included.