quantnado 0.4.4

Dataset generation and peak calling for multi-modal Next-Generation Sequencing data

QuantNado

QuantNado builds per-sample Zarr stores from genomic assays and exposes a unified Python API for region selection, reduction, feature counting, normalisation, PCA, and peak calling.

Installation

pip install quantnado

Requires Python 3.12 or 3.13.

Quick Start

1. Create per-sample stores

quantnado dataset create \
  --sample ATAC_1 \
  --assay ATAC \
  --bamfile /data/ATAC_1.bam \
  --output-dir dataset \
  --chromsizes hg38.chrom.sizes

quantnado dataset create \
  --sample H3K27ac_1 \
  --assay ChIP \
  --bamfile /data/H3K27ac_1.bam \
  --ip H3K27ac \
  --output-dir dataset

quantnado dataset create \
  --sample METH_1 \
  --assay METH \
  --bamfile /data/METH_1.bam \
  --methylation_file /data/METH_1.bedGraph \
  --output-dir dataset

quantnado dataset create \
  --sample SNP_1 \
  --assay SNP \
  --vcf_file /data/SNP_1.vcf.gz \
  --output-dir dataset

This writes one .zarr store per sample into dataset/.

For quick test builds, you can either use the default test chromosomes:

quantnado dataset create \
  --sample ATAC_1 \
  --assay ATAC \
  --bamfile /data/ATAC_1.bam \
  --output-dir dataset \
  --test

or provide an explicit list:

quantnado dataset create \
  --sample ATAC_1 \
  --assay ATAC \
  --bamfile /data/ATAC_1.bam \
  --output-dir dataset \
  --test-chrom chr21 \
  --test-chrom chr9

2. Open the dataset in Python

from quantnado import QuantNado

qn = QuantNado.open("dataset/")

print(qn.sample_names)
print(qn.assays)
print(qn.array_keys)
print(qn.info)

3. Run common analyses

# Select a genomic region
region = qn.sel("chr1", 1_000_000, 1_010_000)

# Reduce signal over intervals
promoters = qn.reduce(
    intervals_path="promoters.bed",
    reduction="mean",
    modality="coverage",
)

# Quantify stored signal over genes
gene_signal, gene_meta = qn.quantify_signal(
    gtf_file="genes.gtf",
    feature_type="gene",
    assay="RNA",
    modality="coverage",
)

# Count features using the current signal backend
counts, features = qn.count_features(
    gtf_file="genes.gtf",
    feature_type="gene",
    engine="signal",
    assay="RNA",
)

# PCA on reduced signal
pca_obj, pca_result = qn.pca(promoters["mean"], n_components=10)

4. Optionally combine stores

quantnado dataset combine \
  --stores dataset/ATAC_1.zarr dataset/H3K27ac_1.zarr dataset/METH_1.zarr dataset/SNP_1.zarr \
  --output dataset/combined.zarr

You can open either dataset/ or dataset/combined.zarr with the same API.

CLI

QuantNado installs a quantnado command with two main workflows.

dataset create

Creates one per-sample Zarr store from direct assay inputs.

quantnado dataset create \
  --sample RNA_1 \
  --assay RNA \
  --bamfile /data/RNA_1.bam \
  --stranded R \
  --output-dir dataset

Supported assays: ATAC, ChIP, RNA, CUT&TAG, METH, SNP, MCC.

dataset combine

Combines per-sample stores into one multi-sample store.

quantnado dataset combine \
  --stores dataset/ATAC_1.zarr dataset/RNA_1.zarr dataset/METH_1.zarr \
  --output dataset/combined.zarr

call-peaks

Calls peaks directly from a QuantNado dataset.

quantnado call-peaks \
  --zarr dataset/combined.zarr \
  --method quantile \
  --assay atac \
  --output-dir peaks/

Available methods: quantile, seacr, and lanceotron.

Python API

The main entry points are:

Object / function	Purpose
QuantNado.open(path)	Open a directory of per-sample stores or a combined .zarr
QuantNado.combine(src, output)	Combine per-sample stores into one multi-sample store
QuantNadoDataset(path)	Lower-level analysis object used by the facade
create_dataset(...)	Build a single per-sample store programmatically
metadata_from_seqnado(...)	Generate a QuantNado metadata table from a SeqNado project

Common analysis methods on QuantNado / QuantNadoDataset:

Method	Purpose
.sel(chrom, start, end, ...)	Extract a genomic region as xr.Dataset
.reduce(...)	Summarise signal over BED/GTF intervals
.quantify_signal(...)	Quantify stored signal over features
.count_features(...)	Count features via the selected engine (signal today, bam planned)
.extract(...)	Bin signal around promoters, genes, transcripts, or exons
.normalise(...)	Apply CPM/RPKM/TPM normalisation
.group_by(...), .subset(...), .info	Notebook-friendly sample grouping, filtering, and dataset summaries
.pca(...)	Run PCA on reduced or selected signal
.metaplot(...), .tornadoplot(...), .heatmap(...), .correlate(...)	Visualisation helpers

Documentation

Full documentation is available at milne-group.github.io/QuantNado.

License

GNU GPL v3.0. See LICENSE.