Workflows¶

This page documents all workflows in the pipeline.

Contents¶

run_pipeline
(entry) (entry point)
TCR_EPITOPE_PIPELINE
NMF_VAE_PIPELINE
BATCH_EFFECT_ASSESSMENTS_PIPELINE
INGEST_EXPORT_PIPELINE
TCR_MIL_PIPELINE
MAKE_TCR_VECTOR_DATABASE_PIPELINE
INGEST_TABULATE_PIPELINE
INTEGRATION_PIPELINE
GEX_MIL_PIPELINE

run_pipeline¶

Defined in main.nf:22

Inputs¶

Name	Description
`<none>`	-

Outputs¶

Name	Description
`<none>`	-

(entry)¶

Entry workflow

Defined in main.nf:125

TCR_EPITOPE_PIPELINE¶

Defined in workflows/tcr_epitope_pipeline.nf:55

Inputs¶

Name	Description
`samplesheet`	-

Outputs¶

Name	Description
`rds`	-
`tcr_rds`	-
`tcr_metadata`	-
`merged_tcr_meta`	-
`clone_embeddings`	-
`binding_scores`	-
`cell_binding_scores`	-
`clone_metadata`	-
`annotated_rds`	-

NMF_VAE_PIPELINE¶

Defined in workflows/nmf_vae.nf:76

NMF-VAE pipeline entrypoint. Downloads Seurat objects, exports raw counts, merges them into a joint .h5ad matrix, and trains an NMF-VAE with signed graph Laplacian regularization (ARCHS4 correlation) to produce shared gene programs. This workflow runs on CPU or GPU depending on the executor profile.

Inputs¶

Name	Description
`samplesheet`	-

Outputs¶

Name	Description
`counts`	-
`merged_h5ad`	-
`genes_file`	-
`latent_z`	-
`decoder_w`	-
`loss_history`	-
`loss_plot`	-
`model_checkpoint`	-

BATCH_EFFECT_ASSESSMENTS_PIPELINE¶

Defined in workflows/batch_effect_assessments_pipeline.nf:74

Inputs¶

Name	Description
`samplesheet`	-

Outputs¶

Name	Description
`prep`	-
`summaries`	-
`run_summary`	-
`ingested`	-

INGEST_EXPORT_PIPELINE¶

Defined in workflows/ingest_export.nf:69

Download Seurat objects and export them as 10x-like count directories. This workflow is intended for fast local or CI validation when users want count matrices without running cross-species harmonization or scMODAL.

Inputs¶

Name	Description
`samplesheet`	-

Outputs¶

Name	Description
`rds`	-
`counts`	-

TCR_MIL_PIPELINE¶

Defined in workflows/tcr_mil_pipeline.nf:98

Inputs¶

Name	Description
`samplesheet`	-

Outputs¶

Name	Description
`rds`	-
`tcr_rds`	-
`tcr_metadata`	-
`merged_tcr_meta`	-
`tcr_model`	-
`tcr_history`	-
`tcr_predictions`	-
`tcr_importance`	-

MAKE_TCR_VECTOR_DATABASE_PIPELINE¶

Defined in workflows/make_tcr_vector_database_pipeline.nf:68

Inputs¶

Name	Description
`samplesheet`	-

Outputs¶

Name	Description
`vectordb`	-

INGEST_TABULATE_PIPELINE¶

Defined in workflows/ingest_tabulate.nf:77

Metadata-only workflow entrypoint. Downloads cell-level metadata tables and aggregates them into a subject-level summary table suitable for cohort QC and downstream analysis. Supports LabKey (output_file_id), URL-based (public RDS), local-file mode, and direct metadata CSV mode (metadata_path).

Inputs¶

Name	Description
`samplesheet`	-

Outputs¶

Name	Description
`metadata`	-
`subject_table`	-

INTEGRATION_PIPELINE¶

Defined in workflows/integration_pipeline.nf:72

Integration pipeline entrypoint. Downloads Seurat objects, exports raw counts, harmonizes genes across species, and trains scMODAL to create a shared latent embedding. This workflow is GPU-backed and intended for SLURM execution.

Inputs¶

Name	Description
`samplesheet`	-

Outputs¶

Name	Description
`rds`	-
`counts`	-
`harmonized`	-
`model`	-

GEX_MIL_PIPELINE¶

Defined in workflows/gex_mil_pipeline.nf:87

Inputs¶

Name	Description
`samplesheet`	-

Outputs¶

Name	Description
`rds`	-
`counts`	-
`merged_gex`	-
`cell_meta`	-
`scvi_model`	-
`mil_model`	-
`config`	-
`metrics`	-
`predictions`	-
`attention`	-

This pipeline was built with Nextflow. Documentation generated by nf-docs v0.4.0 on 2026-06-03 23:27:50 UTC.