Node reference

Functional Activity nodes.

Expression-based activity inference with decoupler: PROGENy scores 14 signalling pathways from signed downstream footprints, Hallmark scores the 50 MSigDB Hallmark programs, and CollecTRI infers transcription-factor activity from curated signed regulons. Start with sample-level pseudobulk and review target coverage; per-cell or per-spot scores are optional exploratory outputs. They answer a different question from DEG-list ORA or GSEA and get their own report section.

03 nodesCategory · activity

Priors are installed, not fetched

Each node needs its prior network (PROGENy, Hallmark, CollecTRI) installed once in the Resource Center, with the licence terms accepted. The version and checksum used are recorded in the result’s provenance; a run never reaches the network. Human priors are what ship in the catalogue today.

PROGENy Activity

progenyActivity

Infer the activity of 14 signalling pathways from signed downstream expression footprints. Uses a pinned PROGENy prior and a multivariate linear model (MLM). Sample-level pseudobulk is recommended for comparisons; optional cell or spot scores are intended for exploratory maps. This is expression-based activity inference, not DEG-list pathway enrichment.

Node LibraryscRNA-seqBulk RNA-seq

InAnnData expression matrix; raw counts layer plus a biological-replicate column recommended

OutAnnData (uns["bf_activity_v1"] sample scores, coverage and provenance; optional observation scores)

Parameters (11)

Name	Type	Default	Description
`analysis_level`	select	`sample`	"sample" aggregates biological replicates; "cell" computes exploratory cell/spot scores.
`sample_key`	obs column	—	Biological replicate column required for sample-level pseudobulk.
`condition_key`	obs column	—	Optional condition metadata retained for replicate-level comparison.
`organism`	select	`human`	Reference organism. Confirm auto-detection before running.
`matrix_source`	select	`auto`	Choose X, raw, or a layer. Auto prefers raw counts for sample aggregation.
`layer`	layer	—	Layer used when matrix_source is "layer".
`store_observation_scores`	checkbox	`false`	Store bounded cell/spot scores; disabled by default to limit output size.
`progeny_top`	select	`100`	Top footprint genes per pathway. Review coverage before increasing to 500 for targeted panels.
`min_targets`	number	`5`	Minimum measured targets required to retain a pathway.
`selected_sources`	text	—	Optional comma-separated pathway names. Empty scores every pathway passing coverage.
`max_output_mb`	number	`256`	Maximum in-memory observation-score output before the run is rejected or externalized.

Hallmark Activity

hallmarkActivity

Score 50 MSigDB Hallmark biological programs from the expression matrix. Uses a pinned Hallmark gene-set snapshot. Sample profiles use ULM; cell/spot profiles use rank-based AUCell. The two estimators have different scales and must not be compared numerically.

Node LibraryscRNA-seqBulk RNA-seq

InAnnData expression matrix; raw counts layer plus a biological-replicate column recommended

OutAnnData (uns["bf_activity_v1"] sample scores, coverage and provenance; optional observation scores)

Parameters (10)

Name	Type	Default	Description
`analysis_level`	select	`sample`	"sample" uses ULM on pseudobulk; "cell" uses AUCell.
`sample_key`	obs column	—	Biological replicate column required for sample-level pseudobulk.
`condition_key`	obs column	—	Optional condition metadata retained for replicate-level comparison.
`organism`	select	`human`	Reference organism. Confirm auto-detection before running.
`matrix_source`	select	`auto`	Choose X, raw, or a layer. Auto prefers raw counts for sample aggregation.
`layer`	layer	—	Layer used when matrix_source is "layer".
`store_observation_scores`	checkbox	`false`	Store bounded cell/spot AUCell scores for exploratory display.
`min_targets`	number	`5`	Minimum measured members required to retain a Hallmark set.
`selected_sources`	text	—	Optional comma-separated Hallmark names. Empty scores every program passing coverage.
`max_output_mb`	number	`256`	Maximum in-memory observation-score output before the run is rejected or externalized.

CollecTRI TF Activity

collectriActivity

Infer transcription-factor activity from curated, signed TF-target regulons. Uses a pinned CollecTRI network with a univariate linear model (ULM). TF activity reflects coordinated target expression and is not the same as expression of the TF gene itself. Select TFs or use sample mode for large datasets.

Node LibraryscRNA-seqBulk RNA-seq

InAnnData expression matrix; raw counts layer plus a biological-replicate column recommended

OutAnnData (uns["bf_activity_v1"] sample scores, coverage and provenance; optional observation scores)

Parameters (10)

Name	Type	Default	Description
`analysis_level`	select	`sample`	"sample" aggregates replicates; "cell" computes bounded exploratory scores.
`sample_key`	obs column	—	Biological replicate column required for sample-level pseudobulk.
`condition_key`	obs column	—	Optional condition metadata retained for replicate-level comparison.
`organism`	select	`human`	Reference organism. Confirm auto-detection before running.
`matrix_source`	select	`auto`	Choose X, raw, or a layer. Auto prefers raw counts for sample aggregation.
`layer`	layer	—	Layer used when matrix_source is "layer".
`store_observation_scores`	checkbox	`false`	Store bounded cell/spot scores; disabled by default because hundreds of TFs can be large.
`min_targets`	number	`15`	Minimum measured targets required to retain a TF regulon.
`selected_sources`	text	—	Optional comma-separated TF list to limit compute and output.
`max_output_mb`	number	`256`	Maximum in-memory observation-score output before the run is rejected or externalized.

Pathway

Cell-Cell Communication