FABIAN-variant

Standalone Version

Contents

Downloads

FABIAN-variant is primarily a web application. The standalone version supports high-throughput analysis, integration into local pipelines, and the processing of sensitive data that cannot be sent to an external server. It is distributed as a self-contained executable with no additional dependencies.

Note: Downloads for the updated standalone version will be available again by 10 July 2026.

Last updated: 7 July 2026.

The application is written in C++11 and has no special hardware requirements. Processing time scales linearly with the number of variants and selected models. TFFMs take longer to evaluate than PWMs.

Command-line interface

The standalone tool supports two modes:

Filters and TF selection

The <filters-and-TFs> argument is a quoted string containing source filters, model filters, and the TF selection. Filters must come before TF names. The standalone binary embeds prediction models from openly licensed sources.

Option Description
Optional source filters (combine any number)
--JASPAR-TFFM JASPAR 2024 TFFMs
--JASPAR-PWM JASPAR 2026 PWM models
--HOCOMOCO HOCOMOCO v14 models
--SwissRegulon SwissRegulon models
--SELEX SELEX collection models (subset)
Optional filters
--human-models Only use models derived from human data
--mouse-models Only use models derived from mouse data
--no-heterodimers Exclude heterodimer TFs (e.g. JUND::FOS)
TF selection (one required, must come last)
--all Analyse all transcription factors
TF1 TF2 ... Analyse specific TFs (space-separated, must be last)

Example commands

Genome data

Genome data is only required when processing VCF files with --file. The download scripts fetch FASTA files for all supported assemblies and save them to data/hg38, data/hg19, etc. Alternatively, chromosome files can be downloaded manually from:

Only the assembly to which the VCF file was aligned is required.

Input formats

Output format

Output is a tab-separated file with per-model predictions, identical to the downloadable results from the web version:

variant tf model_id database ref_score alt_score start_ref end_ref start_alt end_alt strand_ref strand_alt prediction score

See the documentation for column descriptions.

Contact

For questions about the standalone version, please email robin.steinhaus (at) bih-charite.de or submit a ticket via email using this link.