API

The stable public entry point is REVISEPipeline. It resolves revise/revise.yaml profiles, applies runtime and IO overrides, and routes each run to the appropriate backend strategy.

Start with Quick Start for runnable examples and Architecture for the full lifecycle. This page is the reference surface for classes and extension points.

Current API architecture.

from revise.framework import REVISEPipeline

pipeline = REVISEPipeline()
svc = pipeline.run(
    profile="application_sc",
    runtime_overrides={"platform": "iST", "confounding": "segmentation"},
    io_overrides={
        "data_root": "raw_data/Real_application",
        "output_root": "output/sc_SVC_case",
        "sample_name": "P2CRC",
        "st_file": "Xenium.h5ad",
        "sc_ref_file": "adata_sc_all_reanno.h5ad",
    },
    set_overrides=["sc.select_ct=T"],
)

Pipeline

`revise.framework.REVISEPipeline`	Unified orchestration API for all REVISE tasks and modes.
`revise.recon.context.PipelineContext`
`revise.recon.pipeline.UnifiedReconstructionPipeline`
`revise.svc.SVC`	Canonical result carrier for all unified REVISE runs.

Facade Helpers

The facade helpers are thin wrappers around REVISEPipeline.run for callers that want named sp-SVC or sc-SVC functions while still using the unified config surface.

`revise.recon.facade.sp_svc`
`revise.recon.facade.sc_svc`

Configuration

revise/revise.yaml is the external configuration and routing surface. revise.config.runner_conf contains internal runner contracts used by backend adapters and compatibility notebooks.

`revise.config.runner_conf.ApplicationSpConf`
`revise.config.runner_conf.ApplicationScConf`
`revise.config.runner_conf.ApplicationScSrConf`
`revise.config.runner_conf.BenchmarkSrConf`
`revise.config.runner_conf.BenchmarkSegConf`
`revise.config.runner_conf.BenchmarkImputeConf`

Analysis Services

Analysis services consume the unified SVC result carrier and provide notebook-compatible downstream helpers.

`revise.analysis.ScSVCAnalysisService`	Downstream sc-SVC analysis that consumes unified SVC only.
`revise.analysis.SpSVCAnalysisService`	Basic downstream metrics for sp-SVC outputs.
`revise.analysis.compute_metric`	Compute per-gene expression reconstruction metrics.
`revise.analysis.compute_clustering_metrics`	Compute ARI/NMI from predicted and reference labels in adata.obs.

Registries and Plugins

These classes are the extension points used by the unified backend.

`revise.backend.registry.StrategyRegistry`
`revise.backend.registry.PluginRegistry`
`revise.backend.contracts.LocalRefinementStrategy`	Task-level reconstruction strategy contract.
`revise.backend.contracts.PlatformAdapter`
`revise.backend.contracts.CFStrategy`
`revise.backend.contracts.OTSolver`

Backend Compatibility Runners

These classes are kept for notebooks, parity checks, and low-level debugging. New application and benchmark code should prefer REVISEPipeline or the root wrapper scripts.

`revise.backend.runners.sp_svc_application.SpSVC`	sp-SVC class for application usage.
`revise.backend.runners.sc_svc_application.ScSVC`	sc-SVC class for application usage.
`revise.backend.runners.sc_svc_sr_application.ScSVCSr`	sc-SVC super-resolution for application usage.
`revise.backend.runners.sp_svc_benchmark.SpSVC`	sp-SVC class for benchmark CFs: segmentation/bin2cell.
`revise.backend.runners.sc_svc_sr_benchmark.ScSVCSr`	sc-SVC super-resolution for benchmark CFs: spot size/ batch effect.
`revise.backend.runners.sc_svc_impute_benchmark.ScSVCImpute`	Single-cell SVC imputation for benchmark CFs: gene panel/gene dropout.