A Field Map of the Major Databases

By the end of this lesson you will hold a single mental picture of the whole field: which database owns which kind of biological data, and where our worked example — the gene BRCA1 — lives inside each one. That map is the thing you will lean on for the rest of the course.

Organize the field by data type, not by brand

In The Two Hubs: NCBI Entrez and EMBL-EBI you met the two organizations that host most of these resources. But when you sit down to actually find something, the useful question is rarely "which company runs this?" — it is "what type of data do I need?" A DNA sequence, a curated protein, a 3D structure, and a research paper are four different things, and each has a database (or a handful) that specializes in it.

So the map below is organized by data type. Learn this shape once and you will always know where to start, even for a database this course never names.

• Sequence — the raw and curated letters of DNA, RNA, and protein.
• Genome — a whole assembled genome with its genes laid out along chromosomes.
• Protein — a curated knowledge record about one protein: its function, features, and links.
• Structure — the experimentally solved or predicted 3D shape of a molecule.
• Variation — the differences between individuals, and what they mean clinically.
• Pathway — how genes and proteins wire together into biological processes.
• Expression — which genes are switched on, where, and how much.
• Literature — the papers that describe all of the above.

The BRCA1 thread

From here on, one gene runs through every module: BRCA1 (breast cancer type 1 susceptibility gene). It is a real, heavily studied human tumor-suppressor gene, which makes it perfect for our purpose — nearly every database has a rich record for it.

Here is the payoff of the whole course in one table. The same molecule wears a different accession number — a stable, permanent ID for one database record — in each place it lives. A prefix like NM_, ENSG, or P is not decoration; it tells you which database issued the ID and often what kind of molecule it points at.

Notice the two 672s. NCBI Gene uses 672 as its record ID, and KEGG reuses that same number as hsa:672 (the hsa prefix means Homo sapiens). That is not a coincidence — it is a deliberate cross-reference, and spotting patterns like it is exactly the skill this course builds.

Sequence, structure, and expression, close up

A few of these deserve a sharper definition, because beginners often blur them.

Sequence vs structure is the split people mix up most. A sequence is the one-dimensional string of letters (ATGGAT... for DNA, or MDLSAL... for protein). A structure is the three-dimensional arrangement of atoms in space — the folded shape. BRCA1's sequence lives in RefSeq and Ensembl; its structure lives in the PDB (Protein Data Bank), where 1JNX holds the crystal structure of BRCA1's BRCT repeat region and 1JM7 holds the solution structure of the BRCA1–BARD1 RING heterodimer. Sequence tells you what; structure tells you how it works.

Variation is a fourth thing again. dbSNP and ClinVar do not store BRCA1's reference sequence — they store the differences from it seen across people, and, in ClinVar's case, whether a given change is classified as benign or pathogenic. A clinical question about one specific BRCA1 change routes to those variation databases, not to RefSeq.

Expression archives such as GEO answer yet another question: not what the gene is, but whether it is switched on in a given tissue or condition — same gene, entirely different data type.

Touch the map with one line of code

You do not need any of this memorized to start pulling records. Using Biopython — the Python toolkit you installed in the Biopython setup guide — the same efetch pattern you first tried in the intro course's Fetching a sequence lesson pulls BRCA1's curated mRNA straight from NCBI:

from Bio import Entrez, SeqIO

Entrez.email = "[email protected]"   # NCBI asks who you are
handle = Entrez.efetch(db="nucleotide", id="NM_007294",
                       rettype="gb", retmode="text")
record = SeqIO.read(handle, "genbank")
handle.close()

print(record.id, "-", record.description)
print("length:", len(record.seq), "bp")

That db="nucleotide" plus id="NM_007294" is you using the map: sequence data type → NCBI's nucleotide database → the RefSeq accession from the table above. Swap the database and the ID, and the same shape retrieves a protein, a structure, or a variant.

Where each module picks up the thread

You now have the whole itinerary. Later modules in this course each take one row of the BRCA1 table and open it up in full: the NCBI-run resources claim the Gene and RefSeq columns, the EMBL-EBI resources claim Ensembl, UniProt, and PDB, and dedicated modules pick up pathways, interaction networks, expression, and literature in turn. Keep this table nearby — the rest of the course simply fills it in, one accession at a time.