Normalization¶

Normalization corrects systematic biases so that intensity differences between samples reflect true biological variation rather than technical artifacts.

mokume applies normalization at two levels: run-level (within samples) and sample-level (across samples).

Pipeline Overview¶

graph LR
    A[Raw Features] --> B[Run Normalization]
    B --> C[Peptidoform Aggregation]
    C --> D[Sample Normalization]
    D --> E[Protein Quantification]

    style B fill:#e8eaf6
    style D fill:#e8eaf6

Run-Level Normalization¶

Run normalization (--run-normalization) adjusts for intensity differences between technical replicates within each sample. Applied when technical_replicates > 1.

Method	Description	Formula
`median`	Normalize by median	intensity / median(intensity)
`mean`	Normalize by mean	intensity / mean(intensity)
`max`	Normalize by max	intensity / max(intensity)
`global`	Normalize by sum	intensity / sum(intensity)
`max_min`	Min-max scaling	(intensity - min) / (max - min)
`iqr`	Interquartile range	Uses IQR for scaling
`none`	No normalization	—

mokume features2proteins -p data.parquet -o out.csv \
    --run-normalization median

Sample-Level Normalization¶

Sample normalization (--sample-normalization) adjusts for systematic differences across samples. These methods fall into two categories:

Per-Sample Methods¶

Applied during data loading, one sample at a time:

Method	Description
`globalMedian`	Divides each sample by its median, normalized to the global median
`conditionMedian`	Same as globalMedian but within each experimental condition
`none`	No normalization

Dataset-Level Methods¶

Applied after all samples are loaded, operating on the complete dataset:

Method	Description
`hierarchical`	DirectLFQ-style hierarchical clustering normalization
`tmm`	Trimmed Mean of M-values (Robinson & Oshlack, 2010)

When to use hierarchical normalization

Use --sample-normalization hierarchical when you want DirectLFQ-style normalization combined with a different quantification method (e.g., iBAQ). This gives you the normalization quality of DirectLFQ with the quantification approach of your choice.

Global Median¶

The default method. For each sample, computes:

$$\text{normalized} = \frac{\text{intensity}}{\text{sample_median} / \text{global_median}}$$

This ensures all samples have comparable median intensities.

Hierarchical Normalization¶

Uses the DirectLFQ hierarchical clustering approach (Ammar et al., 2023) implemented natively in mokume:

Convert to log2 scale
Align samples using variance-guided pairwise normalization
Convert back to linear scale

You can optionally specify a set of proteins to use for normalization:

mokume features2proteins -p data.parquet -o out.csv \
    --sample-normalization hierarchical \
    --normalization-proteins housekeeping_proteins.txt

TMM Normalization¶

Trimmed Mean of M-values computes normalization factors robust to composition bias from highly abundant proteins. Based on Robinson & Oshlack (2010).

mokume features2proteins -p data.parquet -o out.csv \
    --sample-normalization tmm

DirectLFQ Mode¶

DirectLFQ handles its own normalization

When using --quant-method directlfq, mokume delegates all processing (normalization + quantification) to the DirectLFQ package. The --run-normalization and --sample-normalization options are ignored.