Data analysis as well as base calling was performed with the Illumina instrument software, yielding fastq output files. Further data analysis was based and modified according to Keppel et al. (2020). To remove PCR amplification artifacts, sequencing data was collapsed for each sample. The processed fastq files were mapped to accession NC_003450.3 as C. glutamicum reference genome. This was done using Bowtie2 with the following parameters: --ignore-quals --local --very-sensitive-local --rfg 9,5 --rdg 9,5 --score-min L,40,1.2 -k 8 --no-unal --no-mixed --threads 8 -I 40 -X 800 (Langmead and Salzberg, 2012; Langmead et al., 2019). The genomic coverage was convoluted with second order Gaussian kernel. The kernel was truncated at 4 sigmas and expanded to the expected peak width. The expected peak width was predicted using the following procedure: (i) All peaks higher than 3 mean coverage were detected. (ii) Points at which coverage dropped below half of the maximal peak height were detected and the distance between those was set as peak width. (iii) The estimated peak width was fixed equal to the median peak width. Convolution profiles were scanned to allow identification of the regions where first derivative changes from positive to negative. Each of these regions was determined as a potential peak with an assigned convolution score (convolution with second order Gaussian kernel centered at the peak position). These filtered peaks were normalized for inter-sample comparisons and the sum of coverages of all detected peaks was negated from the total genomic coverage. This difference was used as normalization coefficient and was divided by peak intensities. 

The custom-developed software adapted in this study for ChAP-Seq analysis is publicly available at GitHub repository under the link https://github.com/afilipch/afp.