\@writefile{lof}{\contentsline{figure}{\numberline{A.1}{\ignorespaces\textbf{Predictive power of SMiLE-seq :}\textbf{A} binding models were derived de novo from HT-SELEX 1st cycle data using the HMM discovery method (labelled HT-SELEX cycle 1 HMM) and their performances were assessed using the AUC-ROC. AUC-ROC values for the corresponding TF models derived from SMiLe-seq data (labelled SMiLE-seq) and reported by Jolma and colleagues (labelled HT-SELEX reported matrices, \cite{jolma_dna-binding_2013}) are also displayed. \textbf{B} the predictive performances of CEBPb, CTCF and TCF7 binding models were assessed using subsets of binding sites of decreasing affinities. Inside each peak list, the peaks were ranked by score and subsets of 500 peaks were selected. Peaks 1-500 have the highest affinity, then peaks 501-1000, and so on. The boxplots indicate the distribution of AUC-ROC obtained over all available peak-lists.\relax}}{71}{figure.caption.37}}
\newlabel{suppl_smileseq_auc_2}{{A.1}{71}{\textbf{Predictive power of SMiLE-seq :}\textbf{A} binding models were derived de novo from HT-SELEX 1st cycle data using the HMM discovery method (labelled HT-SELEX cycle 1 HMM) and their performances were assessed using the AUC-ROC. AUC-ROC values for the corresponding TF models derived from SMiLe-seq data (labelled SMiLE-seq) and reported by Jolma and colleagues (labelled HT-SELEX reported matrices, \cite{jolma_dna-binding_2013}) are also displayed. \textbf{B} the predictive performances of CEBPb, CTCF and TCF7 binding models were assessed using subsets of binding sites of decreasing affinities. Inside each peak list, the peaks were ranked by score and subsets of 500 peaks were selected. Peaks 1-500 have the highest affinity, then peaks 501-1000, and so on. The boxplots indicate the distribution of AUC-ROC obtained over all available peak-lists.\relax}{figure.caption.37}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.2}{\ignorespaces\textbf{Chromatine architectures around CTCF binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}}{72}{figure.caption.38}}
\newlabel{suppl_encode_peaks_em_ctcf}{{A.2}{72}{\textbf{Chromatine architectures around CTCF binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}{figure.caption.38}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.3}{\ignorespaces\textbf{Chromatine architectures around NRF1 binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}}{73}{figure.caption.39}}
\newlabel{suppl_encode_peaks_em_nrf1}{{A.3}{73}{\textbf{Chromatine architectures around NRF1 binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}{figure.caption.39}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.4}{\ignorespaces\textbf{Chromatine architectures around cFOS binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}}{74}{figure.caption.40}}
\newlabel{suppl_encode_peaks_em_cfos}{{A.4}{74}{\textbf{Chromatine architectures around cFOS binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}{figure.caption.40}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.5}{\ignorespaces\textbf{Chromatine architectures around max binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}}{75}{figure.caption.41}}
\newlabel{suppl_encode_peaks_em_max}{{A.5}{75}{\textbf{Chromatine architectures around max binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}{figure.caption.41}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.6}{\ignorespaces\textbf{Chromatine architectures around BRCA1 binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}}{76}{figure.caption.42}}
\newlabel{suppl_encode_peaks_em_brca1}{{A.6}{76}{\textbf{Chromatine architectures around BRCA1 binding sites} discovered using ChIPPartitioning. The partition was done with respect to the MNase reads (red), +/- 1kb around the peaks, in bins of 10bp, that were allowed to be shifted and flipped. DNaseI (blue), TSS density (violet) and sequence conservation (green) were realigned according to MNase classification and overlaid. The y-axis scale represent the proportion of the highest signal for each chromatin pattern. The first row contains the aggregated signal over all sites. The number of binding sites (peaks) is indicated in parenthesis. The following rows contains the 4 classes discovered. Their overall probability is indicated atop of the class signal, on the right. The y-axis indicates the min/max signal for all densities.\relax}{figure.caption.42}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.7}{\ignorespaces\textbf{Nucleosome occupancy around CTCF peaks } measured by MNase-seq, in bins of 10bp. The nucleosome depleted region is displayed in blue.\relax}}{77}{figure.caption.43}}
\newlabel{suppl_encode_peaks_ctcf_ndr}{{A.7}{77}{\textbf{Nucleosome occupancy around CTCF peaks } measured by MNase-seq, in bins of 10bp. The nucleosome depleted region is displayed in blue.\relax}{figure.caption.43}{}}
\citation{ou_motifstack_2018}
\citation{ou_motifstack_2018}
\@writefile{lof}{\contentsline{figure}{\numberline{A.8}{\ignorespaces\textbf{Open chromatin classes around SP1 motifs :} EMRead was run without shifing (+/- 10bp) but with flipping to identify different classes of footprints around 15'883 SP1 motifs. The aggregation signal around the 6 different classes found are shown by decreasing class probability. The open chromatin patterns are displayed in red, the nucleosomes are displayed in blue. The aggregated DNA sequence is displayed as a logo. The y-axis ranges from the minimum to the maximum signal observed. For the DNA logo, this corresponds to 0 and 2 bits respectively.\relax}}{78}{figure.caption.44}}
\newlabel{suppl_emread_sp1_noshift_flip}{{A.8}{78}{\textbf{Open chromatin classes around SP1 motifs :} EMRead was run without shifing (+/- 10bp) but with flipping to identify different classes of footprints around 15'883 SP1 motifs. The aggregation signal around the 6 different classes found are shown by decreasing class probability. The open chromatin patterns are displayed in red, the nucleosomes are displayed in blue. The aggregated DNA sequence is displayed as a logo. The y-axis ranges from the minimum to the maximum signal observed. For the DNA logo, this corresponds to 0 and 2 bits respectively.\relax}{figure.caption.44}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.9}{\ignorespaces\textbf{Open chromatin classes around SP1 motifs :} EMRead was run with shifing (+/- 10bp) flipping to identify different classes of footprints around 15'883 SP1 motifs. The aggregation signal around the 6 different classes found are shown by decreasing class probability. The open chromatin patterns are displayed in red, the nucleosomes are displayed in blue. The aggregated DNA sequence is displayed as a logo. The y-axis ranges from the minimum to the maximum signal observed. For the DNA logo, this corresponds to 0 and 2 bits respectively.\relax}}{78}{figure.caption.45}}
\newlabel{suppl_emread_sp1_shift_flip}{{A.9}{78}{\textbf{Open chromatin classes around SP1 motifs :} EMRead was run with shifing (+/- 10bp) flipping to identify different classes of footprints around 15'883 SP1 motifs. The aggregation signal around the 6 different classes found are shown by decreasing class probability. The open chromatin patterns are displayed in red, the nucleosomes are displayed in blue. The aggregated DNA sequence is displayed as a logo. The y-axis ranges from the minimum to the maximum signal observed. For the DNA logo, this corresponds to 0 and 2 bits respectively.\relax}{figure.caption.45}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.10}{\ignorespaces\textbf{Simulated data motifs :} motifs used for the data generation (labeled "True motif") and the best scoring - based on the AUC - partition motifs (labeled "Found motif"). The partition with EMSequence was run such that it was searching for motifs of 11bp, slightly longer than those used for the data generation. "RC" stands for reverse complement. The motifs tree and alignment was build using the motifStack R package \citep{ou_motifstack_2018}.\relax}}{79}{figure.caption.46}}
\newlabel{suppl_atac_seq_emseq_best_motifs}{{A.10}{79}{\textbf{Simulated data motifs :} motifs used for the data generation (labeled "True motif") and the best scoring - based on the AUC - partition motifs (labeled "Found motif"). The partition with EMSequence was run such that it was searching for motifs of 11bp, slightly longer than those used for the data generation. "RC" stands for reverse complement. The motifs tree and alignment was build using the motifStack R package \citep{ou_motifstack_2018}.\relax}{figure.caption.46}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.11}{\ignorespaces\textbf{SP1 motifs :} partition of 15'883 801bp sequences centered on a SP1 binding site using EMSequence. These sequences were classified by EMSequence to search for 10 different 30bp long motifs ($801-30=771$ of shifting freedom). The optimization was run for 20 iterations. The different classes are ordered by decreasing overall probability. Arrows atop of the motifs indicates head-to-tail arrangements of SP1 motifs.\relax}}{80}{figure.caption.47}}
\newlabel{suppl_emseq_sp1_10class}{{A.11}{80}{\textbf{SP1 motifs :} partition of 15'883 801bp sequences centered on a SP1 binding site using EMSequence. These sequences were classified by EMSequence to search for 10 different 30bp long motifs ($801-30=771$ of shifting freedom). The optimization was run for 20 iterations. The different classes are ordered by decreasing overall probability. Arrows atop of the motifs indicates head-to-tail arrangements of SP1 motifs.\relax}{figure.caption.47}{}}
\@writefile{lof}{\contentsline{figure}{\numberline{A.12}{\ignorespaces\textbf{SP1 motifs :} partition of 15'883 801bp sequences centered on a SP1 binding site using EMSequence. These sequences were classified by EMSequence to search for 10 different 30bp long motifs ($801-30=771$ of shifting freedom). The optimization was run for 20 iterations. The different classes are ordered by decreasing overall probability. Arrows atop of the motifs indicates head-to-tail arrangements of SP1 motifs.\relax}}{81}{figure.caption.48}}
\newlabel{suppl_emseq_sp1_10class}{{A.12}{81}{\textbf{SP1 motifs :} partition of 15'883 801bp sequences centered on a SP1 binding site using EMSequence. These sequences were classified by EMSequence to search for 10 different 30bp long motifs ($801-30=771$ of shifting freedom). The optimization was run for 20 iterations. The different classes are ordered by decreasing overall probability. Arrows atop of the motifs indicates head-to-tail arrangements of SP1 motifs.\relax}{figure.caption.48}{}}