When genetic databases are worked to their full depth

Many diseases are complex. They do not follow simple Mendelian inheritance, and they cannot be explained by the presence or absence of a single gene.

Patients with the same genetic variant may show very different clinical outcomes. Some develop disease, others do not. Some experience severe adverse drug reactions, others tolerate the same medication without problems.

This variability is not a failure of genetics. It is a consequence of how genetic data are analysed.

Genome-Wide Association Studies (GWAS) analyse very large cohorts and test genetic variants one at a time. Each variant is evaluated independently to see whether it is statistically associated with a phenotype.

This approach has strengths:

• it scales to millions of individuals
• it detects average, population-level effects
• it is well standardised and widely used

GWAS focuses on marginal effects: the effect of one variant while all others are averaged out.

By design, GWAS cannot fully capture situations where:

• a genetic effect appears only when two or more variants are present simultaneously
• individual variants show little or no effect on their own
• the relevant information lies in specific combinations, not in single markers

In these cases, increasing sample size does not solve the problem. The limitation is not statistical power, but the additive model itself.

GWAS does not fail; it simply answers a different question.

Epistasis refers to gene–gene interaction: situations where the effect of one genetic variant depends on the presence of another.

In epistatic models:

• combinations of variants are treated as entities in their own right
• logical relations such as AND / OR matter
• a disease or adverse reaction may occur only when all required components are present

Here, the informative unit is not the single SNP, but the joint genotype.

This perspective aligns more closely with biological mechanisms and with individual clinical outcomes.

A genetic database can be analysed at different depths:

• analysing single variants extracts partial information
• analysing interactions extracts additional, non-redundant information

When interactions are not studied, part of the information contained in the data remains unexplored.

This is not a philosophical statement. It is an informational one.

AMSAFIS does not perform GWAS.

This is not due to lack of data, tools, or expertise. It is a deliberate methodological choice.

When genetic data are available, and when the objective is to explain diseases or adverse drug reactions, working at the maximum level of information is essential.

For this reason, AMSAFIS focuses on:

• gene–gene interactions
• non-linear models
• explicit combination rules
• explanation of mechanisms, not only prediction

When a genetic database has been analysed, an important question remains: Have genetic interactions been studied?

If the answer is no, then the database has not been explored to its full depth.

Understanding this distinction helps patients, families, and associations better evaluate what genetic studies can — and cannot — tell them about disease, risk, and treatment response.

This wiki is independent from commercial or institutional positioning. Its purpose is to explain the limits and possibilities of genetic analysis in clear terms.