Human genome databases make biological information searchable, but the records do not all answer the same question. One resource describes a gene, another supplies a reference sequence, and another collects observations or interpretations of variants. A useful search begins by identifying which kind of information you need. Otherwise, a familiar gene name or an alarming annotation can distract from the context that gives the record meaning.
Geneomix is an educational information resource. Our overview of human genome databases explains the main resource types. This article offers a practical reading sequence: identify the record, establish the reference, examine the annotation, and trace the evidence.
Choose a resource for the question
For a gene's name, location, products, and associated literature, a gene-centered resource is a useful starting point. NCBI Gene brings together several of these elements and connects them to other databases. A gene page provides an organized entry point into the evidence; it does not mean that every linked study reaches the same conclusion.
Reference sequence collections serve another purpose. NCBI RefSeq supplies genomic, transcript, and protein sequences that researchers can use as common reference objects. Genome browsers, including Ensembl and NCBI Genome Data Viewer, place genes and other annotations along genomic coordinates. They help answer where a feature lies and what else is nearby.
Variant resources focus on differences. dbSNP supplies identifiers and information for short genetic variation. ClinVar archives submitted classifications of variants and their supporting information. Population resources such as gnomAD summarize how often alleles were observed in their datasets. A single investigation may use several resources because location, frequency, and interpretation are separate questions.
Confirm exactly what the identifier represents
A gene symbol is convenient to read, but names and aliases can change. A database identifier provides a more specific way to identify the record. Check the organism as well: related genes in different species can share similar names. The combination of organism, resource, and identifier reduces ambiguity.
Ensembl distinguishes identifiers for genes, transcripts, and proteins. RefSeq also assigns accessions to different types of sequences. A transcript accession is not interchangeable with the gene's identifier because one gene may produce several transcripts. A version suffix can matter when a sequence or annotation changes.
An rs identifier refers to a dbSNP reference variant record. The number is an identifier, not a genomic coordinate, an evidence grade, or a statement that the variant causes disease. Copy it accurately, then read the alleles and mappings within the record. Do not infer the meaning of a finding from the presence of an identifier alone.
Establish the reference assembly before comparing positions
A coordinate describes a position on a particular reference sequence. Human records commonly refer to named assemblies such as GRCh37 or GRCh38. Updates to reference sequence organization can change where a region is located in that coordinate system. Matching only the chromosome and position can therefore compare different things when assemblies differ.
NCBI assigns accessions to genome assemblies, with version numbers tracking changes under its versioning policy. Recording the assembly accession and version is more precise than saving a screenshot that only shows a chromosome label. For most introductory reading, at least record the assembly name and preserve the complete source description.
Suppose two reports describe a variant on chromosome 7 at different positions. Before concluding that they disagree, check whether they used the same assembly, reference sequence, and allele representation. Coordinate conversion can help relate assemblies, but a converted position still needs checking. The surrounding sequence and the actual change provide additional context.
Read genome browser tracks as layered annotations
A browser commonly displays a coordinate ruler with separate tracks for genes, transcripts, variants, and other features. Each track adds a particular dataset or annotation layer. Horizontal overlap shows that features occupy overlapping genomic positions; it does not by itself prove a functional relationship.
Start with the scale. A view spanning a chromosome supports different observations from a view spanning one exon. Then identify the track's source, release, and display settings. Filters can hide features, and the absence of a mark from the current view may simply reflect which data are being shown.
Gene models also have direction. A gene can be represented on either DNA strand, so its transcript runs in a defined orientation relative to the reference. This helps explain why a report may describe a change using different letters when it refers to the complementary strand. Compare complete descriptions rather than isolated letters.
Keep transcript context beside a consequence label
Variant annotations often describe a change as intronic, synonymous, missense, or affecting a splice site. These labels express a relationship between a variant and an annotated feature. A missense consequence indicates an amino acid substitution for a particular protein-coding transcript; it does not independently establish a health effect.
Because a gene can have several transcripts, one genomic change can receive different consequence labels in different transcript contexts. It may lie within an exon in one transcript and outside that exon in another. Preserve the transcript identifier and version when taking notes, especially if comparing a DNA change with a protein change.
Ensembl's documentation on calculated variant consequences explains how these annotations are assigned. Treat an effect prediction as one form of evidence. A computational score, an experimental observation, and a submitted clinical classification have different origins and should be read accordingly.
Interpret frequencies with their denominator
Allele frequency describes how often an allele was observed among the chromosome copies evaluated at that position in a dataset. It is not automatically the percentage of people carrying the allele, because a person may contribute more than one copy. The allele count and number of evaluated alleles help explain what a displayed frequency means.
A frequency also depends on the dataset's composition and the quality of observation at that location. Read group labels, sample context, and filters before comparing values. An allele absent from a database is absent from that dataset's reported observations; that alone does not establish that it is unique, harmful, or absent from the wider population.
Follow classifications to their evidence
ClinVar collects submissions from laboratories, researchers, and other groups. Its records can show agreement or conflicting classifications, and its review status communicates information about the submissions and review process. Read the condition, classification type, submitter, evaluation date, and supporting evidence together.
A review indicator is not a score for the severity of a condition. Likewise, a classification about one variant-condition relationship should not be generalized to every variant in the same gene. An unfamiliar gene can contain many different variants with different consequences. The record's scope is as important as its headline label.
Database information can change as submissions and evidence develop. If comparing a current record with an older report, distinguish a change in the scientific interpretation from a change in the underlying sequence description. Our guide to human genome testing places these interpretation questions within the wider testing process.
Public reference data and individual data have different access rules
Public gene and variant pages do not provide unrestricted access to every participant's genome. For example, dbGaP offers public study descriptions and other summary information while controlling access to individual research data. A publicly visible study record can therefore describe a dataset that requires an approved access process.
For learning how a browser or database works, published identifiers and public example records are usually sufficient. Uploading an individual's genetic file to an analysis service is a separate action with its own data handling terms. Understanding a public record does not require assuming that private genetic information must be shared.
Keep a compact record of what you learned
Useful notes contain the database name, date accessed, organism, identifier, assembly, sequence or transcript version, and the specific claim being investigated. Add the evidence source and unresolved questions. This small amount of context makes the search reproducible and helps explain apparent disagreements later.
If a record remains confusing, return to the underlying question. Our guide to genes, chromosomes, and variation explains the biology, while the comparison of genotyping and sequencing clarifies how observations are produced. Read database entries as structured evidence with a defined scope, and each layer becomes easier to use.
