Differential Abundance Results

After the proteomics result table has been inspected and the sample metadata/comparison design has been documented, the next step is to understand the differential abundance results.

In a results-first proteomics workflow, differential abundance results are often already present in the table received from a proteomics facility, collaborator, or vendor.

These results usually describe how protein abundance changes between two conditions.

For example:

treated_vs_control

may compare a treated condition against a control condition.

The differential abundance table should tell the analyst:

Which protein changed?
        ↓
In which comparison?
        ↓
In which direction?
        ↓
How large was the change?
        ↓
How statistically supported was the change?

This chapter focuses on understanding those result columns before formal quality control, filtering, ranking, enrichment analysis, and interpretation.

Starting Point

This chapter builds on the example files introduced earlier:

data/example/example-proteomics-results.csv
data/example/example-sample-metadata.csv
data/example/example-comparison-design.csv

The main result table is:

data/example/example-proteomics-results.csv

A minimal differential abundance table may look like:

protein_id,gene_symbol,protein_name,log2fc,p_value,adjusted_p_value,comparison
P12345,GENE1,Example protein 1,2.31,0.00001,0.0004,treated_vs_control
Q67890,GENE2,Example protein 2,-1.84,0.00008,0.0021,treated_vs_control
A11111,GENE3,Example protein 3,0.72,0.041,0.1200,treated_vs_control

What Differential Abundance Means

Differential abundance refers to a measured difference in protein abundance between comparison groups.

In this guide, the comparison is represented by:

treated_vs_control

The comparison design defines:

numerator condition   = treated
denominator condition = control

Therefore:

positive log2fc → higher abundance in treated
negative log2fc → lower abundance in treated

or equivalently:

negative log2fc → higher abundance in control

This interpretation depends on the comparison design and should never be guessed.

Core Differential Abundance Columns

A practical results-first table should contain columns that identify the protein, describe the comparison, and report the magnitude and significance of change.

Core columns include:

protein_id
gene_symbol
protein_name
log2fc
p_value
adjusted_p_value
comparison

Each column has a distinct role.

Protein Identifier

The protein identifier column links each row to a protein or protein group.

Example:

protein_id
P12345
Q67890
A11111

Depending on the source, this column may also be called:

Accession
UniProt
UniProt_ID
Protein_ID
Majority protein IDs
Protein.Group

Protein identifiers are important because they support downstream annotation, enrichment analysis, and STRING network analysis.

Gene Symbol

The gene symbol column provides a readable biological label.

Example:

gene_symbol
GENE1
GENE2
GENE3

Gene symbols are often easier to interpret than protein accessions, but they are not always unique across species or databases.

For enrichment analysis, gene symbols may need to be mapped carefully to organism-specific identifiers.

Protein Name

The protein name column provides a descriptive label.

Example:

protein_name
Example protein 1
Example protein 2
Example protein 3

This column helps with interpretation and report writing.

However, protein names can be long, inconsistent, or copied from different databases. They should support interpretation but should not replace stable identifiers.

Log2 Fold Change

The log2fc column describes the direction and magnitude of protein abundance change.

Example:

log2fc
 2.31
-1.84
 0.72

The sign shows direction.

positive log2fc → higher in numerator condition
negative log2fc → lower in numerator condition

The magnitude describes the size of the change.

Examples:

log2fc =  1  → approximately 2-fold higher
log2fc = -1  → approximately 2-fold lower
log2fc =  2  → approximately 4-fold higher
log2fc = -2  → approximately 4-fold lower

Because log2 fold change is symmetric around zero, it is easier to compare increases and decreases.

Raw P-Value

The p_value column reports the statistical evidence for a difference before multiple-testing correction.

Example:

p_value
0.00001
0.00008
0.041

A small p-value suggests that the observed difference is unlikely under the statistical model used.

However, proteomics experiments often test many proteins at the same time. Because of this, raw p-values should usually not be used alone to define significant proteins.

Adjusted P-Value

The adjusted_p_value column accounts for multiple testing.

Example:

adjusted_p_value
0.0004
0.0021
0.1200

This column may also be called:

padj
adj_p_value
adj.P.Val
q_value
FDR

For most downstream filtering, the adjusted p-value or false discovery rate is preferred over the raw p-value.

A common threshold is:

adjusted_p_value <= 0.05

This threshold is not universal. It should be selected based on the study design, sample size, platform, and tolerance for false positives.

Comparison Label

The comparison column links each result to the comparison being tested.

Example:

comparison
treated_vs_control

This is especially important when a table contains multiple contrasts.

For example:

treated_vs_control
disease_vs_healthy
timepoint_2_vs_timepoint_1
drug_A_vs_vehicle

If multiple comparisons are present, filtering and interpretation should usually be performed separately for each comparison.

Direction of Change

Direction should be interpreted using both log2fc and the comparison design.

For the example comparison:

treated_vs_control

with treated as numerator and control as denominator:

log2fc > 0 → higher abundance in treated
log2fc < 0 → lower abundance in treated

This will later support labels such as:

upregulated
downregulated
not_significant

However, the labels “upregulated” and “downregulated” should be used carefully. In proteomics, they usually mean higher or lower abundance in the numerator condition, not necessarily transcriptional regulation.

Differential Abundance Is Not Yet Biological Interpretation

A differential abundance table is an analytical result, not a complete biological conclusion.

For example:

protein_id = P12345
log2fc = 2.31
adjusted_p_value = 0.0004

This tells us that the protein changed strongly and with statistical support.

It does not yet explain:

What pathway is involved?
What process is affected?
Is the protein part of a network?
Does the change support the study hypothesis?
Is the change biologically expected?

Those questions are addressed in later chapters through annotation, enrichment analysis, STRING network analysis, visualization, and biological interpretation.

Common Issues in Differential Abundance Tables

Differential abundance tables may contain issues that need to be checked before filtering.

Common issues include:

• missing protein identifiers
• duplicated protein identifiers
• missing gene symbols
• missing or non-numeric fold-change values
• missing p-values or adjusted p-values
• adjusted p-values reported under unclear column names
• comparison labels that do not match the comparison design
• fold-change direction that is not documented
• mixed protein groups or multiple accessions in one row
• contaminant or reverse-sequence entries
• low-confidence protein identifications

These issues are handled more formally in the next chapter on quality control of result tables.

Recommended Minimal Result Table

For this results-first workflow, a recommended minimal table is:

protein_id
gene_symbol
protein_name
log2fc
p_value
adjusted_p_value
comparison

Additional useful columns may include:

control_mean_intensity
treated_mean_intensity
peptide_count
unique_peptides
coverage
functional_category

These additional columns can help with confidence assessment and interpretation, especially when reviewing individual proteins.

Relationship to Later Chapters

This chapter explains what the differential abundance columns mean.

The following chapters use these columns more actively.

03 Differential abundance result columns
        ↓
04 Quality control of result tables
        ↓
05 Differential protein abundance filtering
        ↓
06 Ranking and filtering DEP
        ↓
07 Identifier cleaning and annotation

Chapter 05 will apply thresholds such as:

absolute log2fc >= 1
adjusted_p_value <= 0.05

Chapter 06 will rank proteins for interpretation.

Looking Ahead

The next chapter focuses on quality control of proteomics result tables.

Before filtering proteins, the result table should be checked for structural problems, missing values, duplicated identifiers, and column consistency.

Differential abundance columns
        ↓
Quality control of result tables
        ↓
Filtering and ranking
        ↓
Biological interpretation