Добавил:
Sekretar
kiopkiopkiop18@yandex.ru
t.me/Prokururor I Вовсе не секретарь, но почту проверяю
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз:
Предмет:
Файл:Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_5529_Библиотеки_им_академика_М_И_Перельмана
.pdf
264 Bioinformatics of Autoimmune Diseases
Entrez Programming Utilities (E-utilities) via Biopython to identify publicly available ATAC-Seq
datasets related to systemic lupus in humans. It runs a predened query that targets BioProjects
associated with the SRA while excluding those under controlled access in dbGaP. Users can easily
modify the script to search for other autoimmune or disease-related datasets by replacing the disease name in the query string. After executing the search, the script retrieves the unique BioProject
identiers (IDs), fetches summary metadata, and prints the corresponding BioProject accession
numbers, which can then be used to locate and download the raw sequencing data for further
analysis.
For demonstrating ATAC-Seq data analysis, we will use publicly available raw data from the
NCBI SRA database, specically from BioProject PRJNA1139785, which is linked to GEO Series
GSE273008. This dataset features ATAC-Seq proling of CD4+ T cell subsets, comparing high
CD38 and low CD38 populations, from both healthy individuals and patients with SLE. The study
includes a total of 12 biological samples, with triplicates for each cell type drawn from three healthy
donors and three SLE patients. Sequencing was performed using the Illumina NextSeq 500 platform. The corresponding raw FASTQ les are available through the SRA Run Selector on the GEO
page, making this dataset a valuable resource for practicing ATAC-Seq data processing and analysis
workows.
Table 8.1 presents the metadata, including the NCBI SRA run IDs, the sample condition (SLE or
healthy), and the CD38 expression level of the CD4+ T cells. This table also reects the overall study
design, which will guide the structure of the downstream analysis. To download the FASTQ les
for each sample, rst copy the SRA run IDs into a plain text le named i d s _ AT AC.t x t , with
one ID per line. Then, use the download _ fastq.py Python script (provided as supplementary
material) to download the corresponding FASTQ les into a directory named raw. After the les
are downloaded, you can compress them using the command gzip * to convert all FASTQ les
into .fa st q.gz format for efcient storage and processing.
8.6.2 ATAC-SEQ PIPELINE
The ATAC-Seq pipeline, implemented in the script provided as supplementary material, comprises a
series of modular functions; each designed to perform a specic and essential step in the processing
TABLE 8.1
The Metadata of the ATAC-Seq Raw Data
SRARUNID Condition CD38
SRR29979670 SLE Low
SRR29979671 SLE High
SRR29979672 SLE Low
SRR29979673 SLE High
SRR29979674 SLE Low
SRR29979675 SLE High
SRR29979676 HC Low
SRR29979677 HC High
SRR29979678 HC Low
SRR29979679 HC High
SRR29979680 HC Low
SRR29979681 HC High
Source: Adapted from BioProject PRJNA1139785 Metadata (NCBI SRA).

265 ATAC-Seq for Autoimmune Diseases
and analysis of chromatin accessibility data. The following paragraphs provide a detailed description of each function and its role within the overall workow.
The function is responsible for preparing the reference genome needed for accurate alignment
of ATAC-Seq reads. It begins by checking for the presence of the hg38 genome in a local reference directory. If the genome is not found, the function downloads the compressed FASTA le
from the UCSC Genome Browser, decompresses it, and generates index les required for both and
bowtie2. The command creates a FASTA index (.fai) le used for rapid sequence lookup, while
producing a set of index les that enable efcient alignment of reads. This ensures that the genome
is fully ready for downstream alignment tasks and eliminates the need for manual preparation by
the user.
The function handles the preprocessing of raw sequencing data by removing sequencing adapters
and low-quality bases from each read. It uses the tool, which is optimized for speed and accuracy,
to automatically detect and trim adapter sequences while preserving high-quality portions of the
read. Trimming is essential in ATAC-Seq because transposase activity often introduces adapters
close to the ends of short DNA fragments. Retaining these artifacts can negatively impact mapping
quality and interfere with peak calling. The function reads the compressed FASTQ le for a given
sample, processes it, and writes a new high-quality FASTQ le to the trimmed directory, ready for
alignment.
The function takes the trimmed reads and aligns them to the reference genome using bow-
tie2, a widely used aligner that efciently handles short DNA sequences. The function constructs
a shell pipeline that pipes the aligned reads directly into samtools, which converts them from
Sequence Alignment/Map (SAM) to BAM format and performs sorting. Sorting ensures that the
resulting BAM les are organized by genomic coordinate, which is necessary for indexing and
later peak calling. The resulting BAM les are saved in the aligned directory and represent the
locations of accessible chromatin in the genome, as inferred from the insertion sites of the ATAC
transposase.
The function renes the aligned data by removing reads that align to mitochondrial DNA, which
can represent a substantial portion of ATAC-Seq reads but do not contribute useful information
about nuclear chromatin accessibility. It uses samtools to extract only those reads that align to
nuclear chromosomes and then sorts and saves the ltered data in new BAM les. This function also
creates BAM index les using samtools index, which are required for efcient data loading in
genome browsers and downstream analysis tools. Filtering at this stage improves the signal-to-noise
ratio in the data and ensures that peak calling focuses on relevant genomic regions.
The function uses MACS2 to detect regions of statistically signicant chromatin accessibility from the ltered BAM les. It treats the aligned fragments as signal tracks and scans the
genome for enrichments that exceed the expected background noise. These enrichments are interpreted as open chromatin regions where TFs or regulatory elements may bind. MACS2 is run with
parameters optimized for ATAC-Seq, including --nomodel, --shift, and to accommodate
the unique transposase insertion patterns. The output includes les, which contain genomic coordinates of peaks, their signicance scores, and additional metadata. These peak les are stored
in the peaks directory and can be visualized or used in downstream motif analysis or chromatin
state modeling.
The function serves as the orchestrator of the entire ATAC-Seq workow. It begins by calling to
ensure the genome is available and indexed, then iterates over each sample listed in the metadata
le. For each sample, it sequentially executes the trimming, alignment, ltering, and peak calling
functions, using metadata elds such as SRARUNID and condition to name les and organize
results. This function coordinates the data ow and ensures consistent and reproducible processing
of all samples in the dataset. The nal result of the pipeline is a well-structured set of les for each
sample, including trimmed FASTQ les, aligned BAM les, and MACS2 peak les, which together
provide a high-resolution map of chromatin accessibility suitable for exploratory analysis, visualization, or integrative epigenomic studies.

266 Bioinformatics of Autoimmune Diseases
8.6.3 ATAC-SEQ PEAK ANNOTATION
We will follow a similar approach to annotate ChIP-Seq peaks using HOMER. For instructions on
installing HOMER and obtaining the necessary genome annotation les, please refer to the previous chapter. Ensure that the HOMER directories are included in your system’s PATH by adding the
following line to your le:
PATH="path_to_homer/homer:$PATH"
PATH="path_to_homer/homer/bin:$PATH"
To download the genome annotation for the human genome (hg38), navigate to the directory
containing and run the following command:
perl configureHomer.pl -install hg38
This will install the annotation les required by HOMER for accurate peak annotation.
The annotation program “atac _ peak _ annotation.py” for ATAC-Seq peak les is
designed to integrate the outputs of the ATAC-Seq processing pipeline with HOMER’s powerful
genomic annotation capabilities. The program begins by reading the metadata le, which contains
essential information for identifying each sample, including its unique SRA run ID and experimental condition (e.g., SLE or healthy). This metadata is used to guide the annotation process
by determining the correct le paths for each peak le that was generated during MACS2 peak
calling. The script systematically constructs the expected location of each le within the directory
based on the sample’s ID and condition, ensuring that only samples with existing peak les are
processed.
The core function of the script, annotate _ peaks, wraps around the HOMER tool anno-
tate Pea k s.pl, which is widely used for assigning functional annotations to genomic regions. For
each sample, it builds a command that calls a n not ateP eak s.pl, specifying the MACS2 peak
le as input and the genome build (hg38) as the annotation reference. HOMER uses precompiled
genomic annotation databases to determine the location of each peak relative to known genomic
features, including TSSs, exons, introns, untranslated regions, and intergenic regions. It also identies the nearest gene to each peak and provides the distance to its TSS. This functional context
allows researchers to better interpret which regulatory regions are likely to be involved in gene
regulation, especially in the context of autoimmune diseases or immune cell differentiation.
The output of the annotation function is saved in a dedicated directory, with one annotated
le generated per sample. These les are plain text and contain rich information for each peak,
including genomic coordinates, feature annotation, associated genes, and proximity to regulatory
landmarks. The script ensures that these outputs are consistently named using the sample ID, making it easy to cross-reference results across the pipeline. If a peak le is missing or not found in its
expected location, the script gracefully skips that sample and continues processing the rest.
The run _ annotation function acts as the orchestrator, looping through every sample in
the metadata and calling the annotation function for each. This design enables automated, highthroughput annotation of ATAC-Seq peak les across dozens or hundreds of samples. Because it
leverages HOMER’s built-in genome annotation databases, the program assumes that the appropriate genome (such as hg38) has been installed using HOMER’s conguration tools. The result
is a scalable and reproducible annotation workow that adds critical biological meaning to the
ATAC-Seq peaks, preparing the data for downstream analysis such as differential accessibility,
motif enrichment, or integration with gene expression datasets.
Each column in the ATAC-Seq annotation output table contributes a distinct layer of biological context, and when interpreted collectively, these features can uncover critical insights into the
regulatory architecture that underlies autoimmune diseases. The annotation output les generated
using HOMER are included as supplementary materials. The following section provides an in-depth

267 ATAC-Seq for Autoimmune Diseases
explanation of each annotation column and its biological relevance. The eld descriptions have been
adapted and expanded based on output formats generated by HOMER (Heinz et al., 2010).
PeakID serves as a unique ID for each detected open chromatin region. It allows researchers to
reference, track, and cross-link specic peaks across different analysis stages, including visualization, motif analysis, and correlation with gene expression.
Chr, Start, and End dene the genomic coordinates of the accessible chromatin region. These
locations are essential for mapping peaks to nearby genes and regulatory elements. In autoimmune
diseases, peaks near immune-related loci, such as cytokine receptors or TFs, can point to dysregulated enhancers or promoters.
Strand indicates the DNA strand orientation of the peak region. While ATAC-Seq is generally strand-agnostic in accessibility analysis, strand information becomes relevant when overlapping peaks with annotated transcripts, especially in genes with antisense or bidirectional promoters
potentially involved in immune cell regulation.
Peak Score reects the statistical condence or enrichment level of the peak, typically derived
from MACS2 peak calling. High scores indicate strong accessibility signals, which often correspond to functionally active regulatory regions. In the context of autoimmunity, high-scoring peaks
within immune cell-specic enhancers can signal hyperactive chromatin states contributing to
aberrant gene expression.
Focus Ratio/Region Size quanties the sharpness or focality of the accessibility signal. A high
ratio suggests a narrowly dened, potentially TF-bound region, whereas broader regions may represent nucleosome-depleted zones or chromatin remodeling. Sharp peaks near genes like FOXP3,
IL2RA, or STAT3 may indicate altered TF binding, providing mechanistic insight into immune
dysregulation.
Annotation describes the general genomic feature the peak overlaps, such as “Promoter”,
“Exon”, “Intron”, or “Distal Intergenic”. This contextual information helps prioritize peaks likely
to inuence gene regulation. For example, promoter-associated peaks in CTLA4 or BACH2 could
affect gene expression patterns in T cells or B cells, potentially impairing immune tolerance.
Detailed Annotation offers a more granular description, specifying the relative position (e.g.,
“Promoter (−50bp)” or “Exon 3 of 8”) and can identify exactly where within the gene structure the
chromatin is open. This can be crucial in autoimmune contexts, where alternative promoters or
enhancer switching may lead to pathogenic isoforms or misregulated transcription.
Distance to TSS reports the distance from the peak center to the TSS of the nearest gene. Negative
values indicate upstream promoter regions; positive values indicate downstream regions. In autoimmunity, peaks near key regulatory genes (e.g., within 1 kb upstream of IL2RA or STAT1) can reveal
enhancer–promoter loops or disrupted regulatory architecture driving immune overactivation.
Nearest PromoterID links the peak to a known promoter element, providing a direct mapping
between chromatin accessibility and transcriptional control regions. This is particularly useful for
associating disease-specic peaks with their regulatory context, especially in promoter switching
events observed in SLE or RA.
Entrez ID, Nearest Unigene, Nearest Refseq, and Nearest Ensembl provide standardized
gene IDs across databases. These are critical for integrating ATAC-Seq data with external resources
like gene expression datasets, GWAS hits, or pathway enrichment tools. For example, if a peak near
TNFAIP3 overlaps with a lupus-associated SNP, its Ensembl ID can facilitate downstream quantitative trait loci (eQTL) or chromatin interaction studies.
Gene Name and Gene Alias provide the common and alternate names of the gene nearest to
the peak. These IDs help quickly recognize immunologically relevant genes. Genes like FOXP3,
STAT3, GATA3, or CD38 are well-known regulators of T-cell differentiation, signaling pathways,
or autoimmunity, and accessibility changes near these genes may reveal epigenetic dysregulation
contributing to disease.
Gene Description offers a brief functional summary, which is helpful for researchers unfamiliar with all gene functions. In autoimmune diseases, descriptions such as “Interferon regulatory

268 Bioinformatics of Autoimmune Diseases
factor” (IRF) or “Tumor necrosis factor receptor” immediately ag pathways involving immune
signaling, inammation, or T-cell exhaustion.
Gene Type indicates whether the gene is protein-coding, non-coding RNA, or pseudogene.
This distinction is important because many autoimmune-associated variants occur in non-coding
regions. ATAC-Seq peaks near lncRNAs or enhancer RNAs (eRNAs) may point to regulatory layers beyond coding gene expression, reecting the complexity of immune gene regulation in disease.
Together, HOMER annotation table for ATAC-Seq peaks allows researchers to not only identify
where chromatin is accessible but also interpret how such accessibility maps to genes, regulatory
elements, and immune functions. In autoimmune diseases, integrating this annotation with differential peak analysis, RNA-Seq, and GWAS results can uncover causal regulatory variants, epigenetically misregulated genes, and new therapeutic targets.
8.6.4 ATAC-SEQ PEAK VISUALIZATION
Once the ATAC-Seq data has been processed, peaks called using MACS2, and genomic features
annotated with HOMER, the next essential step is visualization. Visualization serves not only as a
validation tool but also as a way to explore the biological signicance of chromatin accessibility in
different cell types and disease states. In the context of autoimmune diseases, where subtle regulatory changes can drive immune misactivation, visualizing ATAC-Seq data helps reveal patterns of
open chromatin that are otherwise hidden in raw peak les or summary tables. These visual explorations can expose gene regulatory mechanisms, suggest potential disease-associated enhancers,
and help correlate chromatin structure with gene expression patterns.
A primary approach to visualizing ATAC-Seq data is through genome browsers, such as the
IGV and the UCSC Genome Browser. These platforms provide interactive environments where
researchers can examine read alignments, peak calls, and annotations in the context of the genome.
To do this effectively, the ltered and sorted BAM les from the alignment step are loaded into
the browser alongside their Binary Alignment Index (BAI) les. These les display the depth and
distribution of reads across the genome, allowing for an intuitive view of chromatin accessibility.
In IGV, for example, a sharp peak over the promoter of the IL2RA gene in SLE samples (but not in
healthy controls) can signal disease-specic promoter activation. When viewed in conjunction with
the MACS2 .narrowPeak le, which highlights statistically signicant accessible regions, such
observations become more robust and interpretable.
To improve performance and compatibility with large datasets, BAM les can be converted into
coverage tracks in the BigWig format. Tools such as deepTools’ bamCoverage or the UCSC utility
bedGraphToBigWig allow for the generation of normalized read density tracks across the genome.
BigWig les can then be loaded into IGV or UCSC, providing an efcient way to compare multiple
samples. In autoimmune disease studies, loading BigWig tracks from different CD4+ T cell subsets,
such as high CD38 and low CD38, can reveal accessibility differences at loci like STAT1, FOXP3,
or TNFAIP3, which are known to regulate T-cell function and immune tolerance. These differences
may represent epigenetic signatures of immune activation, exhaustion, or dysregulation that are not
evident from gene expression data alone.
Beyond genome browsers, quantitative visualization of ATAC-Seq data across many genomic
regions provides higher-level summaries of regulatory dynamics. Using tools like deepTools,
researchers can create heatmaps and average prole plots centered around dened genomic features
such as TSSs, enhancers, or even specic sets of differentially expressed genes. For example, a
heatmap of ATAC-Seq signal intensities centered on the TSS of interferon-stimulated genes may
show uniformly high accessibility in SLE samples, reecting a global activation of the interferon
response pathway (a hallmark of lupus pathogenesis). Similarly, when autoimmune-associated SNPs
identied by GWAS are mapped onto ATAC-Seq peaks, one can generate plots that reveal whether
these SNPs lie in accessible chromatin in immune cells, thus suggesting functional non-coding
regulatory variants.

269 ATAC-Seq for Autoimmune Diseases
Incorporating motif enrichment and footprinting analysis into visualization can further enhance
interpretation. With tools like HOMER as shown in the previous chapter, TF binding motifs enriched
within peak regions can be identied and visualized alongside accessibility data. For instance,
in RA samples, increased accessibility in distal enhancer regions may coincide with motifs for
IRF5 or T-bet, suggesting these TFs are active in shaping inammatory gene programs. Visualizing
footprint scores along with ATAC-Seq signal can also identify precise binding sites and their differential usage across conditions. This is particularly useful for understanding how transcriptional
regulators control immune cell phenotypes and how these controls are altered in autoimmunity.
Another powerful approach involves integrating ATAC-Seq data with RNA-Seq, ChIP-Seq, or
Hi-C datasets to create multi-modal genome browser sessions or co-visualized plots. For instance,
chromatin accessibility at the CTLA4 promoter can be visualized alongside its expression levels in
CD4+ T cells from both patients and controls. If chromatin is more open and the gene is upregulated
in patient samples, it suggests a direct regulatory effect. These integrative visualizations not only
enhance interpretability but also allow researchers to formulate mechanistic hypotheses about how
gene regulation is disrupted in disease.
In autoimmune disease research, the ability to visualize chromatin accessibility offers a window
into the regulatory circuitry that governs immune cell behavior. Chromatin changes in regulatory
T cells (Tregs), memory T cells, or B cells can now be directly observed at high resolution, highlighting disease-specic regulatory states. Visualization reveals which regions of the genome are
primed for transcription, how that priming differs between healthy and diseased tissues, and which
genes are potentially dysregulated as a result. It transforms abstract peaks into interpretable regulatory elements, enabling researchers to move from data to insight, from regions to functions, and
ultimately, from mechanisms to therapeutic targets.
8.6.5 NUCLEOSOME POSITIONING, TF FOOTPRINTING, DIFFERENTIAL ACCESSIBILITY, AND MOTIFS
One of the unique strengths of ATAC-Seq is its ability to capture not only regions of open chromatin but also the ne-scale architecture of nucleosomes and TF occupancy. Unlike methods that
require cross-linking or immunoprecipitation, ATAC-Seq preserves the native chromatin landscape
through enzymatic cleavage by the Tn5 transposase, which preferentially inserts sequencing adapters into accessible regions of DNA. This enzymatic activity leaves behind a telltale pattern in the
sequencing data that can be leveraged to infer both nucleosome positioning and TF footprinting,
offering a detailed view of chromatin organization and gene regulation.
Nucleosome positioning refers to the identication of the precise locations along the genome
where nucleosomes (DNA–histone complexes that organize and compact DNA) are situated.
Because nucleosomes physically occlude DNA from binding proteins, their placement plays a critical role in determining gene accessibility and regulation. In ATAC-Seq, nucleosome positioning can
be inferred from the fragment length distribution of sequencing reads. When aligned and plotted by
insert size, three characteristic fragment populations often emerge: short fragments (typically less
than 100 bp) arise from NFRs, intermediate fragments (~200 bp) correspond to mono-nucleosomes,
and longer fragments (~400–600 bp) reect the presence of di- and tri-nucleosomes. These patterns are not random but instead reect the regular spacing of nucleosomes around active regulatory elements. For instance, at TSSs, one often observes a nucleosome-depleted region anked by
well-positioned nucleosomes forming a phased, symmetric arrangement. This reects the biological
requirement for TFs and RNA polymerase to access open DNA, while adjacent nucleosomes delineate regulatory boundaries. Computational tools such as TOBIAS, which correct for Tn5 insertion
bias and extract fragment-length signal patterns, can model these distributions to generate highresolution nucleosome maps and visualize how chromatin structure shifts in response to cellular
states or disease conditions.
In autoimmune diseases, where gene regulation is frequently altered not by mutations but by epigenetic mechanisms, nucleosome positioning maps offer insight into immune cell reprogramming.

270 Bioinformatics of Autoimmune Diseases
For example, in SLE, nucleosome repositioning around interferon response elements may reect
sustained transcriptional activation of proinammatory genes. Similarly, in diseases like RA or MS,
altered nucleosome phasing near immune checkpoints or transcriptional repressors such as FOXP3
and BACH2 can indicate disruptions in gene silencing mechanisms, contributing to the breakdown
of immune tolerance. These positional changes provide a complementary layer of information to
peak-based analyses, highlighting how the physical packaging of chromatin inuences gene accessibility and activity.
TF footprinting extends the analytical resolution of ATAC-Seq even further by identifying
specic DNA regions within open chromatin that are protected from Tn5 insertion due to the
binding of regulatory proteins, most notably TFs. When a TF binds to its DNA motif, it shields
a short segment (typically 6–25 bp) from enzymatic cleavage. This leaves a characteristic local
depletion of sequencing reads or “footprint”, anked by areas of increased insertion activity.
These patterns can be computationally detected and interpreted using tools such as TOBIAS,
HI NT-ATAC, or PIQ (Protein interaction quantitation), which correct for enzymatic bias and
align footprints to known DNA motifs, thereby inferring both TF presence and activity even
in the absence of prior ChIP-Seq data. In autoimmune contexts, TF footprinting is particularly
powerful for identifying which regulatory networks are dysregulated in disease-relevant immune
cells. For instance, CD4+ T cells from lupus patients often show increased footprints for IRF5,
T-bet, and STAT1, reecting chronic activation of interferon signaling pathways. Conversely, the
loss of footprints for transcriptional repressors such as FOXP3 or GATA3 in regulatory T cells
may signal the erosion of immune suppression and tolerance. These footprinting patterns help
decode the cell-specic regulatory logic that governs immune responses and reveal which TFs
may be driving disease pathology. They can also distinguish between immune cell subtypes (such
as regulatory, effector, memory, or exhausted T cells) based on their unique chromatin and TF
activity proles.
Together, nucleosome positioning and TF footprinting transform ATAC-Seq data into a multidimensional chromatin map. This goes beyond simply identifying open DNA regions to uncover the
organizational and regulatory forces shaping the genome in each cell type. In autoimmune diseases,
where cellular identity and function are often perturbed through epigenetic reprogramming rather
than coding mutations, these high-resolution analyses provide crucial insight into the molecular
mechanisms driving immune dysregulation. By exposing the interplay between chromatin structure
and transcriptional control, they also offer a roadmap for identifying candidate biomarkers, therapeutic targets, and regulators of immune homeostasis.
The Python program, provided as supplementary material, implements a complete and modular
workow for analyzing ATAC-Seq data using the TOBIAS suite, with a specic focus on understanding chromatin accessibility, TF binding activity, and regulatory differences in the context of
immune-related diseases such as SLE. The pipeline begins by reading a metadata le that contains
information about individual sequencing samples, including their sample IDs and associated experimental conditions. It then sequentially performs Tn5 bias correction, TF footprinting analysis, and
differential binding detection. The program is designed to generate and organize all outputs within
a structured directory, enabling efcient reuse of intermediate and nal data les for downstream
visualization or statistical interpretation.
The input to the pipeline program consists of several key les and directories that provide both
experimental data and reference resources required for the ATAC-Seq analysis. First and foremost, it requires a metadata le in comma-separated values (CSV) format (meta/metadata.
csv), which contains a list of sample IDs (SRARUNID) and corresponding biological conditions,
such as disease status (e.g., “SLE” or “healthy”) and CD38 expression levels. This metadata is
essential for guiding sample-specic processing and for grouping samples in the differential binding analysis. The program also expects aligned BAM les for each sample, named in the format
{sample _ id} _ filtered.bam, stored in the directory. These BAM les are preprocessed
ATAC-Seq alignments, ltered and ready for downstream analysis.

271 ATAC-Seq for Autoimmune Diseases
In addition to sample data, the program requires a reference genome in FASTA format (referen ce/h g38.fa) to guide the bias correction and footprinting steps. A merged peak le (data/
peaks/merged _ peaks.bed), which consolidates accessible regions across all samples, is
also required for focusing the footprinting analysis on biologically meaningful regions (use Python
program to create merged _ peaks.bed). Furthermore, the pipeline uses a motif database in
MEME format (motifs/jaspar.meme) containing known TF binding motifs, which are used
during the differential binding inference to identify enriched regulators (use to download and
prepare jaspar.meme or visit https://jaspar.elixir.no/downloads/ to download it manually and
rename it). All results and intermediate les are written to an output directory (data/tobias/),
which the program creates and manages automatically. These inputs together ensure that the pipeline can operate in a sample-aware, genome-informed, and motif-resolved manner, providing a
high-resolution analysis of chromatin regulation in immune-related conditions.
The pipeline begins with the function, which loads the sample metadata from a CSV le into
a pandas DataFrame. This table serves as the backbone for guiding all downstream analysis, as it
provides the sample IDs required to locate BAM les and to assign each sample to experimental
groups, such as “healthy control” or “SLE”. This metadata is also critical for grouping samples in
the differential binding analysis later in the workow.
The function is the rst major processing step applied to each sample. It takes a ltered BAM
alignment le for each sample and performs Tn5 bias correction using TOBIAS’s ATACorrect command. This correction is essential because the Tn5 transposase enzyme used in ATAC-Seq exhibits
sequence preference when cutting DNA, which can skew the apparent accessibility signal. TOBIAS
models and corrects this bias by comparing observed insertions to an expected background, outputting a bigWig le that reects bias-corrected chromatin accessibility. In this script, careful le and
path management is implemented to ensure that if TOBIAS creates a directory structure around its
outputs, the nal corrected bigWig le is extracted and stored in a predictable location, avoiding
potential misreads by downstream commands.
Following correction, the function calculates the TF footprint scores using TOBIAS’s
FootprintScores command. This step analyzes the corrected bigWig les alongside the genomic
regions specied in a peak le (merged across all samples) and identies ne-grained patterns of
protection from transposase cleavage. These patterns (dips in signal within otherwise accessible
regions) are indicative of TF occupancy, and the resulting footprint bigWig les contain signal
tracks that can be visualized in genome browsers or fed into further statistical analysis.
The function performs one of the most biologically informative steps in the pipeline. It takes
the metadata and groups samples by a user-specied variable, either by condition (e.g., SLE versus
healthy) or by the level of CD38 expression. It then uses the BINDetect command from TOBIAS to
compare footprint signals across these groups. This differential analysis identies TF motifs that
are signicantly more or less bound in one group versus another, offering clues about which regulatory programs are active or dysregulated in disease. BINDetect not only determines the presence
or absence of footprints but also leverages motif scanning and footprint depth to infer changes in
binding activity across conditions.
The bigWig les produced during the TOBIAS pipeline—specically from the ATACorrect and
FootprintScores steps—are essential for interpreting chromatin accessibility and TF binding activity in ATAC-Seq data. These les are dense, compressed binary formats that store continuous signal
data across the genome and are optimized for efcient visualization and querying. They provide
distinct but complementary layers of information that help reveal both global and local features of
chromatin structure and regulatory protein occupancy.
The bigWig le generated by the ATACorrect step, typically named sample _ filtered _
corrected.bw, contains a bias-corrected representation of Tn5 transposase insertions across
the genome. Raw ATAC-Seq data is inherently biased due to the sequence preferences of the Tn5
enzyme, which can lead to misleading peaks of accessibility. The ATACorrect module adjusts
for this bias by modeling the expected background insertion pattern and correcting the observed

272 Bioinformatics of Autoimmune Diseases
signal accordingly. The result is a cleaner and more accurate signal of true chromatin accessibility. To interpret this le, one typically loads it into a genome browser such as IGV, UCSC Genome
Browser, or JBrowse, and views it alongside gene annotations, raw ATAC-Seq reads, or peak calls.
This allows researchers to assess whether certain genomic regions, such as promoters, enhancers, or
insulators, are accessible in the sample of interest. High corrected signal intensities indicate NFRs
that are open to TFs and other regulatory proteins. This corrected signal can also be used for quantitative comparisons of chromatin accessibility across different conditions or cell types.
The FootprintScores step builds upon the corrected signal to detect regions of localized
protection within accessible chromatin, producing a footprint bigWig le (e.g., sample _
fo ot pr i nt s.bw). This le captures ne-scale depletion of insertions, indicative of DNAbound proteins that occlude the Tn5 transposase from accessing their binding sites. Footprints
are typically seen as narrow dips in signal surrounded by higher accessibility, and these patterns
are often centered on known or predicted TF binding motifs. Viewing this footprint signal in a
genome browser can help identify where specic TFs are actively engaged with DNA. When combined with motif annotation les or gene regulatory databases, these footprints can be mapped to
known TF motifs, allowing inference of which regulatory proteins are likely active in the sample.
Moreover, footprints can be compared across conditions (e.g., healthy versus SLE) to identify differential TF occupancy.
Together, the corrected bigWig and footprint bigWig les provide a multidimensional view of
the chromatin landscape. The former highlights general accessibility, while the latter pinpoints the
precise loci where TFs and other DNA-binding proteins are exerting regulatory control. By integrating these les with gene expression data, motif analyses, and disease phenotypes, researchers can
uncover epigenetic dysregulation patterns, identify candidate regulatory elements, and determine
key TFs that may drive disease mechanisms. This is particularly powerful in immunology and
autoimmunity, where gene regulation often changes in the absence of underlying genetic mutations,
and where understanding the chromatin context of immune cells can lead to the discovery of novel
therapeutic targets.
The output of the BINDetect module from the TOBIAS suite provides one of the most biologically insightful layers of ATAC-Seq analysis by integrating motif information with chromatin
accessibility and TF occupancy to infer differential binding events. This step goes beyond simply
identifying where chromatin is open or where TFs are potentially bound; it evaluates whether the
binding of specic TFs changes systematically between experimental conditions, such as between
healthy controls and autoimmune disease samples. The output consists of a series of structured
les and plots, typically located in a designated results directory (e.g., dat a/tobia s/differ-
ential _ condition), and includes motif scores, log-fold change comparisons, volcano plots,
heatmaps, and lists of differentially bound TF motifs.
The core le in the BINDetect output is a table that associates each motif with its predicted
binding score across all samples and compares these scores between groups. It estimates the degree
to which motifs are more or less protected (i.e., bound) in one condition relative to another by calculating differential footprinting scores and associated statistics. These results are interpretable
both statistically and biologically. For example, a signicantly higher footprinting score for a motif
corresponding to the IRF family in lupus samples compared to controls suggests increased IRF
activity in disease, a nding consistent with the known involvement of interferon signaling in SLE.
Similarly, loss of footprints for regulatory T cell–associated factors like FOXP3 or GATA3 could
indicate disrupted immune suppression mechanisms in autoimmune pathology.
Visualization les such as volcano plots and heatmaps allow for immediate identication of
TFs that show the most pronounced differential binding. Volcano plots typically display log2 fold
change in binding on the x-axis and statistical signicance on the y-axis, highlighting TFs with
strong condition-specic enrichment or depletion. Heatmaps, on the other hand, can cluster samples
and motifs based on their footprinting proles, revealing condition-specic binding signatures that
may correlate with immune cell states or disease severity.

273 ATAC-Seq for Autoimmune Diseases
These outputs can be leveraged in several powerful ways. First, they guide hypothesis generation by highlighting candidate TFs that are differentially active in disease versus control. These
candidates can then be validated using orthogonal methods such as ChIP-Seq or perturbation
experiments. Second, BINDetect results can be integrated with RNA-Seq or single-cell expression
data to correlate differential binding with gene expression changes. For example, if a TF shows
increased binding in SLE samples and its target genes are also upregulated, this supports its role
as an upstream regulator of disease-relevant transcriptional programs. Finally, by comparing differential binding across multiple metadata groups (e.g., high CD38 versus low CD38 T cells or
early- versus late-stage disease), researchers can uncover regulatory patterns that mark immune cell
activation, exhaustion, or plasticity, key features of autoimmune pathogenesis.
In the context of autoimmune diseases, BINDetect is particularly valuable because it helps elucidate the epigenetic and transcriptional dysregulation that underlies immune dysfunction. It identies not only which regulatory elements are active but also which TFs are differentially engaged,
providing mechanistic insight into disease processes and uncovering potential molecular targets for
immunomodulatory therapies. By capturing the dynamic nature of chromatin accessibility and TF
binding, BINDetect transforms ATAC-Seq data into actionable biological knowledge.
8.6.6 EMERGING APPLICATIONS AND MULTI-OMICS INTEGRATION
One of the most exciting frontiers in the study of chromatin accessibility and autoimmune diseases
lies in the integration of ATAC-Seq with emerging technologies that offer unprecedented resolution and functional insights. Among these, scATAC-Seq has revolutionized our understanding
of immune cell heterogeneity by enabling the analysis of chromatin accessibility proles at the
level of individual cells. This is particularly relevant in autoimmune diseases, where dysregulation
often involves subtle shifts in the activation states or subpopulation dynamics of immune cells.
For example, in diseases such as lupus or RA, scATAC-Seq can distinguish between naive and
memory T cells, exhausted effector populations, or rare pathogenic subtypes that may be masked in
bulk analyses. By resolving these differences, scATAC-Seq has shed light on lineage commitment,
enhancer activity, and regulatory circuits specic to disease-associated cell states, revealing how
chromatin remodeling underpins immune dysfunction at the most granular level.
Building upon this, spatial epigenomics is emerging as a complementary approach that maps
chromatin accessibility within the anatomical context of tissues affected by autoimmune diseases.
Unlike conventional sequencing methods that dissociate cells from their microenvironment, spatial techniques preserve the architecture of inamed tissues, such as the synovial lining in RA or
salivary glands in Sjögren’s syndrome, allowing researchers to investigate the relationship between
epigenomic states and spatial niches. This is especially crucial in autoimmune pathology, where
immune cell inltration, local cytokine gradients, and stromal interactions can shape gene regulation in highly localized manners. By combining spatial ATAC-Seq with immunouorescence or
in situ hybridization, researchers are beginning to map regulatory element activity with cellular
and tissue-level precision, unveiling epigenetic heterogeneity that correlates with lesion severity,
immune cell localization, or therapy resistance.
The integration of ATAC-Seq with other single-cell omics technologies, such as single-cell RNASeq (scRNA-Seq) and mass spectrometry-based proteomics, provides an even deeper systems-level
view of immune regulation. While scATAC-Seq captures the accessibility landscape that enables or
restricts gene expression, scRNA-Seq measures the actual transcriptional output. When these datasets are integrated, they allow for the inference of gene regulatory networks, identication of coaccessible regions, and pinpointing of active enhancers that drive the expression of immune-related
genes. Furthermore, coupling this information with proteomic data adds a crucial layer of validation by measuring changes in protein abundance and post-translational modications that directly
inuence cell signaling and function. For example, combining these modalities in the context of
T1D could reveal accessible enhancers linked to IL2RA expression, corresponding transcriptional
Соседние файлы в папке Библиотека им академика М.И. Перельмана
