Improvement of a phylogenetic footprinting method for transcription factor binding sites recognition based on the use of bootstrap trials for the analysis of large bacterial genomic data

The rapid development of high-throughput sequencing technologies has led to an explosive accumulation of high-quality bacterial genome sequence data - their number is approaching three million, and this growth continues. This, in turn, provides additional impetus for the development of technologies for more efficient annotation using analytical methods designed to utilize such large-scale genomic data, as well as for achieving new levels of annotation quality. One such analytical approach is phylogenetic footprinting, which aims to identify motifs corresponding to transcription factor binding sites in the promoter regions of bacterial genomes by comparing corresponding sets of regulatory sequences of orthologous genes in related organisms. The continued accumulation of genomic data has served as the basis for further development of this approach. It has been found that an excessive number of sequences in a set analyzed using phylogenetic footprinting only reduces the accuracy of the method, whereas the inclusion of a sequence selection step in the analyzed set based on data on mutual evolutionary distances improves the method's performance. In this paper, we propose and implement a further step in the development of the phylogenetic footprinting method. This step involves multiple runs of the selection step described above to generate distinct subsamples, subsequent pipeline runs for each subsample, and statistical analysis of the results obtained from multiple pipeline runs. The proposed approach, implemented in the MotifsOnFly method, improves the robustness of motif recognition results obtained from multiple pipeline runs. The effectiveness of the MotifsOnFly method is demonstrated using the analysis of the well-annotated promoter of the Escherichia coli OmpW gene.