Scription factors is much larger than the number of modifications and the importance of individual transcription factor is much smaller than importance of modifications. Therefore in the case of TFs the analysis was simpler. We applied an iterative procedure in which 80 per cent of least important TFs were removed from the information system examined in the previous step. When the number of TFs was smaller than 5 the order NSC 697286 single TF was removed.Classification with Random ForestThe classification was performed in two ten-fold cross-validation setups. In both setups the data was split in ten parts. In the first setup each 1/10th of the data set was once set aside as a test set and the remaining 9/10ths of the data set were used to train the classifier. Then the classification error was measured on the test set. The average error from all 10 test sets is then reported. In the second setup the role of the train and test set are reversed – the classifier is trained using 1/10th of the data set and the error is measured using the remaining 9/10ths of the data.Additional materialAdditional file 1: Table S1 ?Detailed ranking of feature importance. For the convenience of the reader, all supplementary information can also be obtained from the supplementary website http://bioputer. mimuw.edu.pl/papers/enhancer_prediction. Additional file 2: Table S2 ?Results of iterative feature removal. For the convenience of the reader, all supplementary information can also be obtained from the supplementary website http://bioputer.mimuw.edu. pl/papers/enhancer_prediction. Additional file 3: Table S3 ?Training set. For the convenience of the reader, all supplementary information can also be obtained from the supplementary website http://bioputer.mimuw.edu.pl/papers/ enhancer_prediction. Additional file 4: Table S4 ?Redfly mesodermal testing set. For the convenience of the reader, all supplementary information can also bePodsiadlo et al. BMC Systems Biology 2013, 7(Suppl 6):S16 http://www.biomedcentral.com/1752-0509/7/S6/SPage 7 ofobtained from the supplementary website http://bioputer.mimuw.edu.pl/ papers/enhancer_prediction. Additional file 5: Table S5 ?Redfly non-specific testing set. For the convenience of the reader, all supplementary information can also be obtained from the supplementary website http://bioputer.mimuw.edu.pl/ papers/enhancer_prediction.8.9.10. 11.Competing interests The authors declare that they have no competing interests. Authors’ contributions BW designed the study; AP prepared all datasets and performed initial classifier training with BNfinder, SVMs and RFs PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/26226583 as well as the functional validation against the Redfly database. Final RF training and feature importance measurements with Boruta package was performed by MW, WP and WR. BW drafted the manuscript based on contributions from all authors. Acknowledgements This work was partially supported by the National Center for Science grant decision numberDEC-2012/05/B/N22/0567 and Foundation for Polish Science within Homing Plus programme co-financed by the European Union?European Regional Development Fund. Declaration The publication fee was covered in full by the National Center for Science grant decision number DEC-2012/05/B/N22/0567. This article has been published as part of BMC Systems Biology Volume 7 Supplement 6, 2013: Selected articles from the 24th International Conference on Genome Informatics (GIW2013). The full contents of the supplement are available online at http://www.biomedcentra.