Since the sequencing depth of the 454 dataset is not very high (2.86×) and contains a eukaryote organism, we also simulated a dataset with higher depth (9×) that included only the prokaryotic species from the dataset.
After a normalization procedure that corrects the read counts taking into account the control sample, the different sequencing depths of the datasets and the nucleosome abundance at each locus, we obtained a joint histone modification level distribution and we applied the k-means clustering algorithm on it (see Supplementary Material for more details).
The sequencing depth of the grouper dataset is ~110-120x, whish is much smaller than the depths of the E. coli and B. cereus datasets in Table 1.
The amount and depth of this dataset enabled us to carry out a systematic analysis of insertions and deletions in proteins inferred from these transcripts by comparing to proteins from a range of other organisms (including a range of metazoans).
Upon initialization and depending on the depth of the HiC dataset, the software splits the contigs into smaller pieces/bins encompassing at least 2 restriction fragments.
For the remaining cell line, NHEK, the sequencing depth of the available dataset was too shallow to reliably call footprints with Wellington, so we extracted only high confidence footprints previously called by a Hidden Markov Model (HMM) algorithm [ 23], which were originally based in the hg18 genome but later lifted over to the hg19 genome.
However, even if a large fraction of transcripts appeared to be detected in this study, the fact that only 21'746 contigs are mapped with 100 or more reads indicates that the sequencing depth of the Illumina dataset alone (53.6M reads) might be insufficient to assemble full length contigs for weakly expressed transcripts.
An in-depth analysis of the dataset validity using the Principal Variables approach was introduced.
Such loci can be avoided by limiting the number of discordant alignments that may be attributed to a putative cluster, as informed by the depth of the input datasets.
