Sovereign 8s.is mirror of standardgalactic/glass-idh
  • R 79.4%
  • Python 12.9%
  • Shell 4.7%
  • Jupyter Notebook 2.9%
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
2026-09-18 16:46:23 -04:00
analysis chore: reorganizing analysis scripts 2026-09-18 15:45:27 -04:00
figures docs: upload scripts batch 3 2026-09-18 15:09:18 -04:00
legacy_codes refactor: reorganize inherited GLASS repo into preprocessing_pipeline/ and legacy_codes/ 2026-05-05 13:53:57 -04:00
preprocessing_pipeline refactor: reorganize inherited GLASS repo into preprocessing_pipeline/ and legacy_codes/ 2026-05-05 13:53:57 -04:00
tables/R chore: update README.md and SETUP.md; remove tables/python 2026-09-18 16:24:08 -04:00
.gitattributes chore: update .gitattributes to reflect new Snakefile location 2026-05-05 14:24:57 -04:00
.gitignore chore: set up new manuscript folder structure with analysis, figures, and tables 2026-05-05 14:22:29 -04:00
LICENSE chore: update LICENSE and README 2026-05-05 14:33:37 -04:00
README.md chore: update README.md and SETUP.md; remove tables/python 2026-09-18 16:24:08 -04:00
SETUP.md chore: update README.md and SETUP.md; remove tables/python 2026-09-18 16:24:08 -04:00

GLASS_logo_072219_08

The GLASS consortium

Overview

The Glioma Longitudinal AnalySiS (GLASS) consortium consists of clinicians, bioinformaticians, and basic science researchers from leading institutions across the world striving to better understand glioma tumor evolution and to expose its therapeutic vulnerabilities. The code in this respository was used to generate the figures and perform the analyses described in our 2025 publication in BioRxiv.

R code used to make each of the figures can be found in the figures/R subdirectory.

Data Release version 2025.

The data analyzed using this code was part of the fourth release of the GLASS dataset which was made available in 2025. This dataset is managed internally using the PostgreSQL database management system. These data are under active curation so future versions will include additional data as well as correct potential errors.

Data Download

The GLASS data can be downloaded from the Tables page here and the Files page here. It is also possible to query the data directly using the the API by using queries. You can read more about that here.

Prior releases

Previous projects and associated repositories are listed below:

Barthel FP, Johnson KC, Varn FS, et al. Longitudinal molecular trajectories of diffuse glioma in adults. Nature. 2019; 576(7785): 112-120. Paper. Code.

Varn FS, Johnson KC, Martinek J, et al. Glioma progression is shaped by genetic evolution and microenvironment interactions. Cell. 2022; 185(12): 2184-2199.e16. Paper. Code.


Repository layout

The repository separates the code behind this manuscript from the standardized GLASS infrastructure it was built on top of. The first three directories below are the former; the last two are the latter.

Directory What it holds
analysis/ Analyses that generate derived results such as copy number, structural variants, mutational signatures
figures/ One script per manuscript figure panel
tables/ One script per manuscript table, mostly cohort description and survival models
legacy_codes/ Code from the prior GLASS publications listed above, retained for provenance
preprocessing_pipeline/ The GLASS workflow that produced the database itself

Environment and database configuration for analysis/, figures/ and tables/ is documented separately in SETUP.md.

analysis/

Derived analyses: everything that computes a result which the figures and tables then read. 23 files in six subdirectories, split by language and by role — R and Python for code you call directly, bash/ and slurm/ for the drivers that orchestrate multi-stage runs, snakemake/ for the cluster workflows.

Subdirectory Files Contents
analysis/R/ 12 Chromothripsis calling (ShatterSeek), ecDNA/amplicon analysis (AmpliconSuite), allele-specific copy number (ASCAT), whole-genome doubling, fraction-of-genome-altered calculation, GISTIC input preparation, copy-number signature exploration, telomere-length normalisation
analysis/SQL/ 4 Query definitions consumed by the scripts and pipelines — CNA extraction, the hypermutant and non-hypermutant dN/dS inputs, GISTIC preparation
analysis/bash/ 2 Two self-contained, resume-aware pipeline drivers: mutational_signature_pipeline.sh (repeat-region filtering → SigProfilerMatrixGenerator/Extractor → Palimpsest deconvolution) and kataegis_pipeline.sh (clustered-mutation / kataegis calling via SigProfilerSimulator + SigProfilerClusters). Both accept --dry-run
analysis/python/ 1 ampsuite_manifest.py, the AmpliconSuite sample-manifest builder called by ampsuite.smk
analysis/slurm/ 1 GLASS-I_SvABA.sbatch, the SvABA structural-variant calling job
analysis/snakemake/ 3 Cluster workflows over BAM files: ampsuite.smk, ascat.smk, sigprofilertoolkit.smk. See the note in SETUP.md §4.3 — their helper-script paths need editing before they run

figures/

One script per figure panel, 28 files. Each is standalone: it opens the GLASS database, builds its own plotting table, and writes the panel.

Subdirectory Files Contents
figures/R/ 27 Cohort description and QC Sankey; swimmer's plot; copy-number heatmap and per-subtype CN segment plots (astrocytoma and oligodendroglioma, primary / recurrence / delta); fraction-of-genome-altered, with and without hypermutants and by grade; chromosomal-instability landscape and CIN markers; IDH mutation figures; mutation burden; mutational-signature boxplots; dN/dS selection; sSNV mutual exclusivity; TMZ cycles; and Kaplan–Meier curves for CNA, drivers (GENIE), hypermutation, ID8 and sSNV burden
figures/python/ 1 The world map of cohort contributing centres

tables/

One script per manuscript table, 16 files, all R.

Group Contents
Cohort description Baseline table, molecular cohort summary, prior-treatment table
Drivers Driver gene table, drivers by prior treatment, IDH mutation table, McNemar test on paired CNA changes
Survival Univariable and multivariable models for three endpoints — overall survival (OS), post-recurrence survival (PRS) and time to recurrence (TTR) — each split into clinical and molecular covariates

Output is formatted with gt and gtsummary.

legacy_codes/

Created separately, outside the scope of this manuscript's analysis code, under the same standardized GLASS pipeline protocols. 333 files of R (308), Python (19) and Julia (5) from the earlier GLASS publications listed under Prior releases above, retained for provenance and reproducibility of those papers.

Organised by topic under legacy_codes/R/: single-cell deconvolution (CIBERSORTx), neoantigen analysis, SNV and CNV processing, manifest handling, mutation timing, figures and tables for the 2019 and 2022 papers, and a Shiny app. The Julia code covers subclonal selection analysis.

This directory is not maintained against the current data release and its dependencies are not part of SETUP.md.

preprocessing_pipeline/

Created separately, through the standardized GLASS consortium pipeline protocols described in the 2025 publication linked above and established in the prior GLASS releases listed under Prior releases. 251 files. This is the workflow that produced the PostgreSQL database the rest of the repository queries — it is not re-run as part of reproducing a figure, and it is not covered by SETUP.md.

Subdirectory Files Contents
sql/ 125 The schema and queries that populate and serve the database, grouped by domain: clinical, variants, snv, cnv, drivers, mut_freq, mut_sig, pyclone, timing, neutrality, neoag, immune, lohhla, expression, heatmap, metadata, set, dndscv, figures
snakemake/ 30 Per-stage workflow modules, driven by the top-level Snakefile
envs/ 44 One conda environment specification per tool — alignment, GATK4, Sequenza, PyClone, Manta, Delly, LUMPY, OptiType, pVACseq, MiXCR, Kallisto, and others
conf/ — Cluster and pipeline configuration
bin/, jar/ — Bundled executables and Java dependencies
dag/, dbm/ — Workflow DAGs and the database schema diagram

Because this directory carries its own environment specifications in envs/, use those rather than the instructions in SETUP.md if you need to re-run any part of it.


Where to start

  • Reproducing a figure or table → read SETUP.md, configure the database connection, then run the single script for that panel.
  • Reproducing a derived analysis → the corresponding script in analysis/R/, or one of the two drivers in analysis/bash/, which accept --dry-run.
  • Understanding how the data was generated → preprocessing_pipeline/ and the methods of the manuscript.
  • Looking for code from the 2019 or 2022 papers → legacy_codes/, or the dedicated repositories linked under Prior releases.