Node reference

ATAC-seq nodes.

scATAC-seq from fragments, a pre-computed peak matrix or a Cell Ranger ATAC MEX directory. ATAC QC reports TSS enrichment, FRiP and fragment-size distribution; Peak Calling uses MACS3 (top-N accessible regions as a fallback); LSI is TF-IDF plus SVD; Chrom Accessibility tests two groups; Motif Analysis scans differential peaks against JASPAR; ATAC Visualization draws 6 diagnostic plot types.

09 nodesCategory · atacin 2 templates

JASPAR and the genome

JASPAR2020 / 2022 / 2024 / 2026 ship inside the installer and are selected in the Resource Center or the node’s database dropdown; the default stays JASPAR2022 so existing analyses reproduce exactly. Scanning sequence under a peak needs a reference genome FASTA that you point the Resource Center at — it is not downloaded for you. Since v0.25.0 Motif Analysis fails loudly when its motif database is missing instead of returning nothing.

Common pitfalls

Neighbors after LSI, not PCA. Set use_rep='X_lsi' in Neighbors; the first LSI component tracks sequencing depth and is usually dropped.

Fragment files can be large. Real fragments.tsv.gz from Cell Ranger ATAC is block-compressed; Load Fragments builds the peak × cell matrix from it and calls peaks when no BED is given, which takes time on a full run.

Load ATAC MEX

loadATACMEX

Cell Ranger ATAC MEX directory (matrix.mtx + peaks.bed + barcodes.tsv).

Node LibraryATAC-seq

OutAnnData (cells × peaks)

Parameters (2)

Name	Type	Default	Description
`directory_path`	directory	—	filtered_peak_bc_matrix/ directory.
`singlecell_csv`	file (.csv)	—	Optional explicit path to Cell Ranger's per-barcode singlecell.csv. Empty = look for it beside the matrix directory and in its parent. An explicitly given file that cannot be imported is an error; an auto-discovered one only adds a warning.

Load Fragments

loadFragments

Build peak×cell matrix from 10X fragments.tsv.gz. Calls peaks if no BED provided.

Node LibraryATAC-seq

OutAnnData (cells × peaks)

Parameters (8)

Name	Type	Default	Description
`fragments_file`	file (.tsv, .bed, .gz)	—	Indexed fragments file (.tbi required).
`peaks_file`	file (.bed, .narrowPeak)	—	Optional peak BED.
`whitelist_file`	file (.tsv, .txt, .csv, .gz)	—	Optional barcode whitelist (barcodes.tsv / .tsv.gz, one barcode per line). Only listed barcodes are loaded. Raw fragment files contain hundreds of thousands of empty droplets, so without a whitelist only min_fragments separates cells from background.
`count_mode`	select	`fragments`	'fragments' counts each fragment once (standard). 'reads' sums the read-support column, which includes PCR duplicates and inflates deeply duplicated libraries; it reproduces the pre-v0.28 behaviour.
`min_fragments`	number	`200`	Min fragments per cell.
`bin_size`	number	`5000`	Fixed genomic bin width (bp) used when peaks_file is empty. Each fragment is assigned by its start coordinate; bins are not inferred from accessibility density.
`max_cells`	number	`50000`	Cap the number of highest-fragment cells loaded. Set 0 for no cap; the 50,000 default protects memory on very deep datasets.
`chunk_rows`	number	`250000`	Rows streamed from the fragments file per batch. Lower values reduce peak RAM without sampling or changing the scientific result.

Load Peak Matrix

loadPeakMatrix

Load pre-computed peak×cell matrix (CSV or H5AD).

Node LibraryATAC-seq

OutAnnData (cells × peaks)

Parameters (3)

Name	Type	Default	Description
`matrix_file`	file (.csv, .tsv, .txt, .h5ad)	—	Path to CSV / H5AD / Matrix-Market peak matrix file.
`transpose`	checkbox	`false`	True if peaks are columns.
`index_col`	number	`0`	Column index in the file holding peak names (chr:start-end). Usually 0.

ATAC QC

atacQC

Per-cell QC — total fragments, peaks, FRIP score.

Node LibraryATAC-seq

InAnnData (ATAC)

OutAnnData (obs total_fragments/n_peaks_detected/frip_approx)

Parameters (4)

Name	Type	Default	Description
`min_total_fragments`	number	`1000`	Minimum total fragments per cell.
`max_total_fragments`	number	`100000`	Maximum fragments per cell (filters potential doublets).
`min_frip`	number	`0.15`	Min FRIP.
`mode`	select	`label`	"label" or "filter".

Peak Calling

peakCalling

Call peaks with MACS3 (or top-N accessible regions fallback).

Node LibraryATAC-seq

InAnnData (ATAC)

OutAnnData (peaks subset)

Parameters (4)

Name	Type	Default	Description
`method`	select	`top_accessible`	"macs3" or "top_accessible".
`keep_top_n`	number	`100000`	Number of top peaks to keep ('top_accessible' method or as a cap on MACS3 output).
`genome_size`	select	`hs`	MACS3 effective genome size. Options: 'hs' (human, 2.7e9), 'mm' (mouse, 1.87e9), 'ce' (C. elegans, 9e7), 'dm' (D. melanogaster, 1.2e8), or a numeric override (e.g. '2.5e9').
`q_value`	number	`0.05`	MACS3 q-value.

LSI

lsi

Latent Semantic Indexing (TF-IDF + randomized SVD). Standard ATAC dim-reduction.

Node LibraryATAC-seq

InAnnData (ATAC, QC'd)

OutAnnData (obsm["X_lsi"])

Parameters (4)

Name	Type	Default	Description
`n_components`	number	`50`	LSI components.
`scale_factor`	number	`10000.0`	TF-IDF scaling constant (Stuart et al. lsiTransform default). Rarely changed.
`log_transform`	checkbox	`true`	Apply log1p to TF-IDF values before SVD (standard LSI recipe). Disable only for debugging.
`drop_first`	checkbox	`true`	Drop the first LSI component (typically correlated with sequencing depth).

Chrom Accessibility

chromAccessibility

Differential chromatin accessibility between two cell groups (t-test + FDR).

Node LibraryATAC-seq

InAnnData (ATAC, with clusters)

OutAnnData (uns["diff_accessibility"])

Parameters (7)

Name	Type	Default	Description
`groupby`	obs column	—	Column in adata.obs with group labels for comparison.
`reference_group`	obs value	—	Reference group name.
`test_group`	obs value	—	Test/comparison group name. Positive log2FC = higher accessibility in test_group.
`test`	select	`wilcoxon`	'wilcoxon' runs scanpy rank_genes_groups on a temporary depth-normalized, log1p sparse copy of X (scATAC convention). 'welch' is the pre-v0.28 Welch t-test on raw counts and ignores sequencing-depth differences.
`alpha`	number	`0.05`	FDR significance threshold.
`log2fc_threshold`	number	`1.0`	Minimum absolute log2 fold-change to mark a peak as significant.
`min_pct`	number	`0.05`	Minimum fraction of cells in either group with the peak open. Filters extremely rare peaks before testing.

Motif Analysis

motifAnalysis

TF motif enrichment in differentially accessible peaks via JASPAR / pyjaspar.

Node LibraryATAC-seq

InAnnData (with diff_accessibility)

OutAnnData (uns["motif_enrichment"])

Parameters (6)

Name	Type	Default	Description
`database`	select	`JASPAR2022`	JASPAR database release (JASPAR2020–JASPAR2026).
`genome_fasta`	file (.fa, .fasta)	—	Genome FASTA required for sequence scanning and motif enrichment. Empty returns database motifs only and does not test peaks. An explicitly missing file is an error.
`tax_group`	select	`vertebrates`	JASPAR CORE taxonomic collection to fetch: vertebrates (human and mouse), plants, fungi, insects, nematodes or urochordates.
`n_top_peaks`	number	`1000`	Number of top differential peaks (by \|log2FC\|/FDR) to scan for motifs.
`max_peaks`	number	`2000`	Upper bound on how many foreground sequences and how many GC-matched background sequences are actually scanned. Replaces the fixed cap of 300 used before v0.28.
`alpha`	number	`0.05`	P-value / FDR threshold for motif enrichment reporting.

ATAC Visualization

atacVisualization

FRIP bar, fragment size, peak accessibility heatmap, ATAC UMAP, differential volcano.

Node LibraryATAC-seq

InAnnData (ATAC)

OutPNG image

Plots (6)frip_barfragment_sizetss_enrichmentpeak_accessibilityumap_atacdifferential

Parameters (3)

Name	Type	Default	Description
`plot_type`	text	`frip_bar`	frip_bar/fragment_size/peak_accessibility/umap_atac/differential.
`color_by`	text	—	Cell color column (UMAP).
`n_top_peaks`	number	`0`	Top peaks in heatmap/volcano.

Methylation