PharmCAT — Before the Report Becomes a Prescribing Decision
Why It Matters
Pharmacogenomic data do not constitute prescribing information based solely on the presence of variants in a VCF file. Allele combinations at multiple loci must be grouped into pharmacogene haplotypes, and diplotypes and functional phenotypes must be inferred before linking them to current guidelines for specific drugs. PharmCAT is an open-source tool that transforms this iterative process into reproducible reports.
The advantage of automation lies in applying the same rules to the same input and version. However, tools cannot autonomously recover variants absent from the input, complex structures, or ambiguous phases. One must read not only the sentences in the report but also which inputs and rules generated them.
From Input to Report
The PharmCAT workflow normalizes genomic positions and observes alleles at defined pharmacogene loci to generate haplotype and diplotype candidates. It then converts the diplotype into a metabolizer or functional phenotype and links it to annotated recommendations, such as those from CPIC. Each step relies on the genome build, reference allele, and allele-definition database.
The meaning of "no variant" differs depending on whether the VCF originates from whole-genome sequencing or a targeted assay. Failure to distinguish between positions where no observation was possible due to insufficient coverage and positions definitively observed as reference may result in the generation of incorrect star alleles.
Complex pharmacogenes
Loci with copy-number variations, deletions, duplications, and hybrid alleles, such as CYP2D6, are difficult to accurately call using a small list of variants. For HLA, high polymorphism, phase, and typing resolution are critical. It is essential to verify how the tool supports these loci and whether the upstream caller provided the necessary structural information.
no-call, multiple possible diplotype possibilities, and incomplete inputs are not failures but significant result states. Automatically substituting the most common allele to conceal them results in an error that appears precise.
Small Example
Even with the same VCF file, one may have the entire pharmacogene detected with high quality, while another may contain only a few hotspots. Although the superficial result of having no additional variants in both files is identical, the first supports reference observation, whereas the latter leaves the possibility of unmeasured regions. The report must convey both the assay range and call completeness.
Versions and Provenance
Pharmacogene allele definitions and clinical guidelines are updated. Reproducible reports include the PharmCAT version, genome build, input preprocessing steps, allele-definition resources, guideline/drug label snapshots, and generation date. When past and current reports differ, first compare these provenance differences.
Because tool updates may necessitate re-interpretation of existing patient results, institutions should define policies for version changes and result dissemination.
Order of Reading the Report
First, assess the certainty of the call and phenotype, as well as unmeasured regions and warnings. Next, determine whether an actual recommendation exists for the gene–drug pair, including its strength and applicability conditions. Finally, make clinical judgments considering concomitant medications, organ function, indications, and local regulatory approvals and policies.
Common Misconceptions
- PharmCAT is not an all-in-one assay that performs sequencing or variant calling itself.
- The appearance of a drug in the report does not necessarily indicate a need for prescription modification.
- A no-call result is distinct from a normal phenotype.
- The automated report and the clinician’s final medication review represent different stages.
Interpretive Caveats
The PharmCAT results are limited by input completeness and the versioned knowledge base. Clinical use requires validated assays, quality control, current guidelines, and the judgment of a qualified professional.
Reading in Context
The clinical scope of gene–drug outcomes is pharmacogenetics, while the technical blind spots of complex loci lead to concepts such as pseudogene mapping and WGS·WES·panel selection.