Guided case study
Module 2 · Biological data and databases · Self-checked
A newborn's screening result leads to a CFTR variant on both copies of the gene. Follow it from the official symbol to the transcript, the protein, its structure, its clinical meaning and the papers behind it.
This case brings together everything in Module 2. The baby is invented for teaching, but the results are the kind a real cystic fibrosis team would see, and the gene, the variant and the identifiers are real. Your job is not to diagnose: the diagnosis is already made. It is to follow one gene and one variant through every kind of database you have met, and to check what each one tells you. Read the whole case file once before you start.
- Patient
- A baby girl, born at full term and feeding well.
- Newborn screening
- Heel-prick blood spot at 5 days: immunoreactive trypsinogen (IRT) raised. Referred to the cystic fibrosis team.
- Sweat test
- Sweat chloride raised, in the range consistent with cystic fibrosis.
- Genetic testing
- Gene: CFTR. Reference: NM_000492. Result: c.1521_1523del (p.Phe508del), homozygous.
- Request
- The family has asked for a clear written summary of the gene and the variant. The team asks you to prepare the database facts behind it.
Terms in the case file
- Newborn screening is a blood test offered to every baby a few days after birth, from a few drops of blood taken from the heel and dried on a card. It looks for a set of rare but serious conditions that do better with early treatment. A screening result is a flag, not a diagnosis.
- IRT (immunoreactive trypsinogen) is a protein made by the pancreas. In babies with cystic fibrosis, blocked ducts in the pancreas push more of it into the blood, so a raised level is a reason to test further.
- A sweat test measures the salt in sweat. The CFTR protein normally helps reabsorb chloride from sweat, so when it does not work, sweat is unusually salty. A sweat chloride of 60 mmol/L or more is consistent with cystic fibrosis.
- Homozygous means both copies of the gene, one from each parent, carry the same variant. Each parent is therefore likely to be a carrier, with one changed copy and no symptoms.
- NM_000492 is a RefSeq accession for an mRNA (Identifiers and cross-references: following one gene across databases). The c. and p. parts are the standard variant notation from Module 1: c. for the coding DNA, p. for the protein. del means deleted.
- You will also meet the older name F508del, or ΔF508, in papers. It means the same variant: phenylalanine (F) 508 deleted. Legacy names like this matter when you search the literature.
Two of these terms come with a common confusion. A raised IRT on its own does not mean cystic fibrosis: it can be raised for other reasons, and many babies with a raised level turn out not to have the condition, which is why the sweat test and the genetic test follow. And "homozygous" in a report describes the two copies of the gene in the child; confirming that each parent is a carrier needs testing of the parents themselves, which the family may or may not choose.
Go deeper: why this one variant is so common
The CFTR gene was identified in 1989, and the first variant found in it was this one. It turned out to be by far the most common cause of cystic fibrosis: in people of northern European ancestry, roughly seven in ten disease-causing CFTR copies carry it. It is less dominant in many other populations, which is one reason screening panels differ between countries.
Its frequency has also shaped treatment. Because the protein is made but folds badly (task 4), drugs can help it. Correctors, such as elexacaftor and tezacaftor, help the protein fold and reach the cell surface; a potentiator, ivacaftor, helps the channel open once it is there. The three-drug combination was approved in the US in 2019 and in Europe in 2020, and is now licensed for people with at least one F508del copy, so for this baby the variant is also a guide to medicine.
How to approach this
Work through the tasks in order, because each one uses the answer before it. Open the databases in a browser as you go and find each fact yourself; the model answers are there to check you, not to replace the search. For each fact, note where you found it and the identifier of the record, as a curator would. The figure below shows the route.
A simple table works well for the notes: one row per fact, with columns for the fact, the database, the record's identifier and the date you looked. It may feel slow, but it is what turns a set of browser tabs into something you can check later and hand to someone else. Databases change between releases, as The main databases warned, so the date tells a reader which version of each record your summary reflects. If two databases seem to disagree, note both values in the table rather than picking one; the disagreement is often the most useful thing to report.
How to read this figure
Follow the arrows: along the top row from left to right, down, then along the bottom row from right to left. The top row is about the gene and its transcript, the bottom row about the protein, its shape and its meaning for patients. The shaded cards mark the start, the official symbol, and the end, the published evidence. Under each name is the task that uses it.
Find the official symbol and gene IDs
The main databases: sequences, genomes, proteins, structures and Identifiers and cross-references: following one gene across databases
Confirm the official gene symbol and full name, then collect the gene's identifiers in HGNC, NCBI Gene and Ensembl, and the chromosome it lies on.
Hint. Start at HGNC, which owns human gene symbols, and search for CFTR. The HGNC record lists the approved name, previous names and aliases, and links out to the other databases, so you can follow the links rather than search again. Check that every record you open is for Homo sapiens.
Starting from the symbol, not a free-text search, matters here. A search for "CF" in a general database returns many unrelated hits, because the letters are used for other genes and terms, while the approved symbol and the HGNC ID point to exactly one gene. Every identifier you collect in this task is a key that the later tasks use.
Check your answer
Symbol CFTR; full name cystic fibrosis transmembrane conductance regulator. HGNC:1884; NCBI Gene ID 1080; Ensembl ENSG00000001626. It lies on chromosome 7. You may also see the alias ABCC7, which reflects its membership of the ABC transporter family; searching by the approved symbol avoids confusion.
Pick the reference transcript
File formats you will meet: FASTA, GenBank, GFF, VCF, PDB, Identifiers and cross-references: following one gene across databases and Reading a gene record and a protein record
The report names NM_000492. What kind of record is this, and why does the version number after the dot matter for reading the variant?
Hint. Recall the RefSeq prefixes from Identifiers and cross-references: following one gene across databases. The c. number counts bases along one particular transcript, starting from the first base of the start codon. Ask what would happen to that count if a different transcript, or a corrected version of the same one, had a different length before position 1521. In NCBI Gene or Ensembl, look for the transcript marked MANE Select.
Check your answer
NM_ marks a curated RefSeq mRNA. A variant's c. position only has meaning on a stated transcript: on another transcript, or another version, the same base can carry a different number. So a full report gives the accession with its version, such as NM_000492 followed by .1, .2 and so on, and you should check that the database record you compare with uses the same one. MANE Select marks the one transcript per gene that NCBI and Ensembl agree on as the default reference, and clinical reports increasingly use it.
Read the variant notation
File formats you will meet: FASTA, GenBank, GFF, VCF, PDB and Reading a gene record and a protein record; Module 1
Explain c.1521_1523del and p.Phe508del in plain words. What kind of change is it, and what happens to the reading frame?
Hint. The underscore joins the first and last positions of a range. Count how many bases are removed and divide by 3, the length of a codon. Then read the p. part: Phe is the three-letter code for phenylalanine. Ask whether the codons after the deletion are still read in the right groups of three.
Check your answer
Bases 1521 to 1523 of the coding sequence are deleted: three bases. Because three is a whole codon's worth, the reading frame is kept: an in-frame deletion. The protein loses one amino acid, phenylalanine 508, and every amino acid after it is unchanged, so the protein is one residue shorter. A detail worth noticing: the three bases straddle the codons for residues 507 and 508, but codon 507 still codes for isoleucine, so the net effect is exactly the loss of Phe508.
Find the protein record and the domain
The main databases: sequences, genomes, proteins, structures and Reading a gene record and a protein record
Find the reviewed UniProt entry for human CFTR. What does the protein do, and in which domain does residue 508 sit?
Hint. A reviewed entry is from Swiss-Prot, the manually curated part of UniProt (The main databases: sequences, genomes, proteins, structures). Read the Function section first, then the "Family & Domains" section, where domains are listed with their start and end positions. Find the one whose range includes 508.
Check your answer
UniProt P13569. CFTR is a channel in the membrane of epithelial cells, such as those lining the airways, gut and sweat glands, that lets chloride and bicarbonate ions through; it belongs to the ABC transporter family. It has two membrane-spanning domains, two nucleotide-binding domains that bind ATP, and a regulatory (R) domain. Residue 508 lies in the first nucleotide-binding domain, NBD1. Losing Phe508 makes the protein fold poorly, so most of it is destroyed inside the cell before it reaches the surface; this is the reason the sweat is salty.
Look for structures
File formats you will meet: FASTA, GenBank, GFF, VCF, PDB and The main databases: sequences, genomes, proteins, structures
Is there a 3D structure of CFTR? Say how you would find experimental and predicted structures, and what you would check before using one.
Hint. The UniProt entry has a Structure section that lists PDB entries and a predicted model. You can also search the PDB by the UniProt accession, and AlphaFold DB the same way. For each experimental entry, open it and read the method, the resolution and which part of the protein was studied.
Check your answer
Yes. Experimental structures exist, including cryo-EM structures of full-length CFTR and crystal structures of NBD1 on its own; there is also a predicted model in AlphaFold DB. Before using one, check: the method and resolution; which residues were included; whether the protein is normal or carries F508del or other engineered changes added to stabilise it; and any bound molecules, such as ATP or drugs. For a predicted model, check the confidence score (pLDDT) per region, and remember it is a model of the normal sequence, not of the variant. Take every PDB ID from the database itself, never from memory or an AI summary.
Clinical significance and the literature
The main databases: sequences, genomes, proteins, structures and Searching and mining the literature
Find the variant in ClinVar and note its classification. Then write a PubMed query for review articles about this variant.
Hint. Search ClinVar for CFTR c.1521_1523del, then read the classification, the condition, the review status and how many submitters agree. For PubMed, build two blocks as in Searching and mining the literature: one for the gene, one for the variant's many names, including the legacy names from the case-file terms.
When you read the ClinVar record, look beyond the headline word. A classification backed by several submitters who agree, or by an expert panel, carries more weight than one lab's report, and the record shows which applies. For the literature, a review is the right place to start for a variant this well known, as Searching and mining the literature suggested; the primary papers it cites are where the original measurements sit.
Check your answer
ClinVar classifies the variant as pathogenic for cystic fibrosis, with many submitters in agreement; read the review status yourself to see how strongly it is reviewed. A suitable query:
("Cystic Fibrosis Transmembrane Conductance Regulator"[mh] OR CFTR[tiab]) AND (F508del[tiab] OR Phe508del[tiab] OR "delta F508"[tiab] OR deltaF508[tiab]) AND review[pt]
The variant block needs all the spellings, because papers from different decades name it differently.
What the case shows
Look back at the six tasks together. One symbol led to a set of identifiers, and the identifiers did the work: each database linked to the next without a fresh search. The variant only made sense once you knew which transcript it was counted on. The protein record turned a three-letter deletion into a reason: a misfolded domain, a missing channel, salty sweat. ClinVar gave the agreed clinical meaning, and the literature search, with all its old names, opened the evidence behind it. No single database told the whole story; together they did, and every fact came with a record you could cite.
The route also shows where the traps were. Had you mixed a mouse record with a human one, used another transcript, or trusted an unreviewed protein entry, each later step would have inherited the error without any warning. Checking the species, the version and the review status at each step is what kept the chain sound. The same route works for any gene: the databases and identifiers stay the same, and only how much each record holds changes.
This module worked one record at a time, in a browser. Real projects need the same files and identifiers in bulk: thousands of FASTA records, a GFF of every gene in a genome, a VCF of millions of variants. Module 3, The command line, starting with The terminal and where to run it, builds on what you have learned here by showing how to open, search, count and filter these files at scale in the terminal, where a few short commands do in seconds what would take days by hand.
Take it further
Repeat the route for a second CFTR variant chosen from ClinVar, this time one classified as a variant of uncertain significance. Compare how much each database can tell you about it.
You will probably find that the gene-level steps look the same, but the variant-level answers thin out: fewer submitters, fewer papers, perhaps conflicting interpretations. That contrast shows how much of what we "know" about a famous variant comes from decades of curated literature, and why a new variant needs evidence before it can be called harmful or harmless.
Further reading · Module 2ClinVar help: classifications and review status. · HGVS Nomenclature: how c. and p. variant names are written. · UniProt help: reading an entry. · PDB-101: an introduction to 3D structures. · All reading for this course
Up next
Lesson 1
The terminal and where to run it