TidyMS2 Batch corrector

A batch corrector plugin for TidyMS2.

Installation

pip install tidyms2-batch-corrector

Usage

This TidyMS2 plugin implements the BatchCorrector, a matrix operator that corrects time-dependent effects on data. It implements the correction described here.

The BatchCorrector can be tuned to fit your experimental design and desired data quality:

• first_n_qc: Controls how many QC samples are used to estimate the intra-batch mean. Use a small value (e.g 1-3) if early QC samples best represent unbiased instrument response.
• frac: Sets the LOESS smoothing parameter. If not specified, an optimal value is estimated for each feature. You can set a fixed value to control the degree of smoothing.
• threshold: Sets the minimum intensity for QC samples to be considered in the correction. Increase this value to ignore low-quality or noisy reference measurements.
• target_groups and ref_groups: Specify which sample groups are corrected and which are used as references. By default, experimental samples and technical QCs are used.
• max_workers: Controls parallelism for batch correction, useful for large datasets.

Import it and use it as a regular matrix operator:

from tidyms2_batch_corrector import BatchCorrector

bc = BatchCorrector(max_workers=4)
bc.apply(matrix)

Below is a minimal example of how to use the BatchCorrector with simulated data:

from tidyms2.core.enums import SampleType
from tidyms2.core.operators.pipeline import Pipeline
from tidyms2.simulation.base import InstrumentResponseSpec
from tidyms2.simulation.lcms import SimulatedLCMSAdductSpec, simulate_data_matrix
from tidyms2.simulation.utils import create_sample_list

from tidyms2_batch_corrector import BatchCorrector

def create_example_matrix():
	"""Create a data matrix with time-dependent variation effects."""
	sample_types = [
		SampleType.TECHNICAL_QC,
		SampleType.SAMPLE,
		SampleType.TECHNICAL_QC,
		SampleType.SAMPLE,
		SampleType.TECHNICAL_QC,
		SampleType.SAMPLE,
		SampleType.TECHNICAL_QC,
	]
	samples = create_sample_list(sample_types, n_batches=3)
	adducts = [
		SimulatedLCMSAdductSpec(
			formula="[C10H20O4]+",
			n_isotopologues=1,
			response=InstrumentResponseSpec(max_sensitivity_loss=0.5, sensitivity_decay=0.25),
		)
	]
	return simulate_data_matrix(adducts, samples)

matrix = create_example_matrix()
matrix.check_status()
pipe = Pipeline("matrix_pipe")
bc = BatchCorrector(id="test-corrector", first_n_qc=1)
pipe.add_operator(bc)
pipe.apply(matrix)