FABIAN-variant Predicting the impact of DNA variants on transcription factor binding.
This is an example of a mutation in the promoter of PIGM which disrupts SP1 binding.
You can either leave the choice of transcription factors as it is (‘All TFs known at position’) or manually choose SP1.
After you are finished, click on ‘Submit.’
1. Enter variant
Variant (VCF-like)
Position must refer to genome build 37.

E.g., 1:160001799G>C or 1:160001798AG>G 
OR
Wild-type sequence
E.g., GGCCCTCACACTCTCCAACCTCATCTCCCTGGTGAGAGGCC
Variant sequence
E.g., TCACACTCTCCAACCTCATCTCCCTGGTGAG
OR
Run a batch query insteadlink
2. Select transcription factors


3. Search for binding sites
Documentation
Simply enter the DNA variant you would like to analyse into the Variant (VCF) field, select one or multiple transcription factors and click on Submit. If you do not know the location but have a wild-type and a variant sequence, you can still enter them. In this case, FABIAN cannot display any known transcription binding sites.
There is detailed documentationlink of FABIAN's output.
We also provide a disease-linked real-life examplelink.
FABIAN currently only works with human transcription factors.
About FABIAN
FABIAN supports more than 1,000 different binding models for 247 human transcription factors. The models were pooled from various publicly accessible data sources, such as CIS-BP, Hocomoco, hPDI, JASPAR and Jolma (2013). 103 transcription factor flexible models (TFFMs) are included. For each binding model, FABIAN constructs a Bayes classifier and reports the likelihood that a variant leads to the gain or loss of a transcription factor binding site. Because of the poor performance of several binding models in detecting experimentally verified sites, a negative or positive predictive value is reported.
Download the underlying data.
FABIAN has been developed in 2020 at Berlin Institute of Health (BIH) and Charité – Universitätsmedizin Berlin by Robin Steinhaus, Peter N. Robinson and Dominik Seelow. FABIAN is an update of the ePOSSUM software.
Imprint / privacy