PSLE-SCI-REALITY-0557
Wait, what? The “50” in N50 is not a score out of 100
A genome report lists two assemblies. Assembly A has N50 = 1,000,000 bases. Assembly B has N50 = 200,000 bases. A learner says, “Assembly A must be 50% accurate, because N50 has fifty in the name. Assembly B is probably less than 50% accurate because its number is smaller.”
The mistake begins before any calculation. N50 is not a percentage accuracy score. It is a summary of how the assembled sequence is broken into pieces. Roughly stated, N50 is the sequence-piece length such that at least half of all assembled bases are contained in contigs or scaffolds that are that length or longer. A bigger N50 usually indicates greater contiguity, but it does not by itself prove higher sequence accuracy, greater biological completeness or fewer assembly errors.
This is a perfect Reality Lab object because the number looks familiar enough to invite the wrong interpretation. PSLE Science reasoning asks you to identify what a quantity actually measures before you use it as evidence.
Quick Answer
No. N50 does not mean 50% accurate or 50% complete. It is a contiguity statistic. Sort the contigs or scaffolds from longest to shortest and keep adding their lengths. When the running total reaches at least half of the total assembly length, the length of the sequence piece that takes you across that halfway point is the N50. To judge an assembly responsibly, also check total assembly size, number of pieces, whether the statistic is contig N50 or scaffold N50, sequence accuracy, missing regions, duplicated sequence, structural errors and independent validation.
The Exact Learner Job
This volume owns one job: how to evaluate the N50 line in a genome-assembly report without converting a contiguity metric into a percentage of accuracy, completeness or truth. It does not own genome assembly algorithms, sequencing chemistry or genetics generally.
For nearby but different evidence objects, use Vol.556 on the human reference genome, Vol.472 on 30× genome coverage, and Vol.499 on Q30 sequencing quality.
Build the Statistic From Scratch
Suppose a fictional genome assembly contains five contigs:
| Contig | Length |
|---|---|
| A | 400,000 bases |
| B | 250,000 bases |
| C | 150,000 bases |
| D | 120,000 bases |
| E | 80,000 bases |
The total assembly length is 1,000,000 bases. Half is 500,000 bases. Begin with the longest contig: 400,000. That is not yet half. Add the second: 400,000 + 250,000 = 650,000. We have now crossed the halfway point. The second contig is 250,000 bases long, so this assembly has N50 = 250,000 bases.
Notice what we did not calculate. We did not ask whether 50% of the bases were correct. We did not count how many genes were present. We did not compare the sequence with a reference. We asked how the total assembled length is distributed among sequence pieces.
Why the Name Contains “50”
The “50” comes from the halfway point in the cumulative assembled length. Related statistics can use different thresholds, such as N25 or N75. The number after N is therefore part of the definition of the cumulative fraction, not a mark awarded to the assembly.
This is a useful general habit in science: a familiar-looking number can have a specialised role. Read the definition before importing an everyday interpretation.
Observed, Claimed and Inferred
| Layer | Example |
|---|---|
| Observed in report | N50 = 1 Mb. |
| Direct meaning | At least half of the assembled bases lie in contigs or scaffolds of length 1 Mb or greater, under that N50 definition. |
| Reasonable comparison | This assembly may be more contiguous than another assembly built from the same target with much smaller N50, if the compared statistic is defined the same way. |
| Unsupported inference | The assembly is 50% accurate. |
| Unsupported inference | The assembly contains 100% of the biological genome correctly. |
Contig N50 and Scaffold N50 Are Not Automatically the Same Job
A contig is a continuous assembled sequence without gaps in that representation. A scaffold may connect contigs using linking evidence while containing unresolved gaps. Therefore a scaffold N50 can be much larger than a contig N50 without meaning that all sequence inside those scaffolds has been continuously read or resolved.
Before comparing two N50 values, check whether both are contig N50 or both are scaffold N50. Comparing a scaffold N50 from one assembly with a contig N50 from another can make the larger number look more impressive while measuring a different level of structure.
A Bigger N50 Can Coexist With Errors
Imagine two distant genomic regions are mistakenly joined into one long contig. That incorrect join can increase contiguity statistics because the assembly now contains a longer piece. The N50 may rise even though the assembly contains a structural error.
This is why assembly evaluation uses more than one metric. Contiguity is useful, but it is only one dimension of quality. The best evidence packet asks multiple questions: Are the pieces long? Are their bases accurate? Are expected regions present? Are sequences duplicated? Are large structures assembled correctly? Do independent data support the joins?
A Smaller N50 Can Still Belong to a Valuable Assembly
Suppose a difficult genome contains many repeated regions. A cautious assembly may leave some parts separate rather than joining them without enough evidence. Its N50 may be lower than an aggressive assembly that makes risky joins. The lower N50 alone does not prove the cautious assembly is worse.
Scientific comparison must therefore ask what trade-off created the number. “Bigger” is not a universal synonym for “better.”
Worked Case 1: Same N50, Very Different Assemblies
Assembly X and Assembly Y both report N50 = 500 kb. X has a total assembly length close to the expected genome size and strong independent validation. Y is missing a large chromosome segment. Do the equal N50 values make the assemblies equally complete? No. N50 does not carry completeness by itself.
Worked Case 2: Higher N50, Worse Base Accuracy
Assembly A has N50 = 2 Mb but many base-level errors. Assembly B has N50 = 800 kb but far fewer base-level errors. Which is “better”? The question is underspecified. A is more contiguous by N50; B may be more accurate at individual bases. The purpose of the analysis determines which properties matter.
Worked Case 3: Scaffold N50 Wins the Beauty Contest
A report advertises scaffold N50 = 20 Mb in large type and places contig N50 = 600 kb in a footnote. A learner compares 20 Mb directly with another project’s contig N50 of 1 Mb and announces the first assembly is twenty times better.
That comparison is invalid because the two numbers refer to different assembly objects. First align the definitions. Then compare like with like.
Worked Case 4: The Total Assembly Size Is Strange
An organism is expected to have a genome near 500 Mb. A report gives N50 = 5 Mb but the total assembly is only 250 Mb. The N50 sounds impressive, but half of an unexpectedly small assembly may still leave much of the biological genome unrepresented. Total size is a necessary context check.
Worked Case 5: A Headline Says “N50 Doubled”
A new method increases N50 from 1 Mb to 2 Mb. What can you safely say? The assembly became more contiguous by that metric. You cannot yet say accuracy doubled, completeness doubled, errors halved or biological usefulness doubled.
Method and Comparison Checks
- Is this contig N50 or scaffold N50?
- What is the total assembly length?
- What is the expected biological genome size, if known?
- How many contigs or scaffolds are present?
- Were tiny pieces filtered before the statistic was calculated?
- Are both assemblies being evaluated with the same rules?
- What independent evidence checks misassemblies?
- What evidence measures base-level accuracy and completeness?
Evidence That Strengthens a “Better Assembly” Claim
- N50 improves while total assembly size remains biologically plausible.
- Independent structural checks find fewer misassemblies.
- Base-level accuracy is maintained or improved.
- Expected genes or conserved regions are recovered more completely.
- Contamination and duplicate sequence are assessed.
- The comparison uses the same N50 definition and filtering rules.
Evidence That Weakens It
- Only N50 is shown.
- Scaffold N50 is compared with contig N50.
- Total assembly size is implausibly small or large.
- Longer pieces contain more unsupported joins.
- Base accuracy declines.
- The headline converts N50 into a percentage score.
Tempting Reasoning That Fails
- “N50 means 50% accurate.” No.
- “N50 means half the genome is missing.” No.
- “A larger N50 always means a better assembly.” Not without other quality checks.
- “Scaffold and contig N50 are interchangeable.” No.
- “Two assemblies with equal N50 are equally correct.” N50 does not establish that.
- “Doubling N50 doubles scientific quality.” Quality has multiple dimensions.
How Far Can the Conclusion Travel?
N50 can support a statement about assembly contiguity. With a fair comparison it can help describe whether one assembly is broken into longer pieces than another. It cannot alone establish correctness, completeness, biological representativeness or fitness for a particular research question.
This is the broader scientific lesson: a metric should carry only the claim it was designed to measure.
PSLE-Style Transfer Case
Two fictional assemblies are reported:
| Assembly P | Assembly Q | |
|---|---|---|
| Contig N50 | 900 kb | 600 kb |
| Expected genome size recovered | 82% | 98% |
| Independent structural errors found | 14 | 3 |
Question 1: Which assembly is more contiguous by N50? Answer: P.
Question 2: Does P therefore have higher overall quality? Answer: Not from N50 alone. Q recovers more of the expected genome and has fewer detected structural errors.
Question 3: What is the strongest conclusion? Answer: P has the larger contig N50, while Q is stronger on the two other reported quality dimensions. More evidence is needed for an overall judgement.
Delayed Independent Return: P-H-Q
- P — Pieces: Are these contigs or scaffolds, and how long are they?
- H — Halfway: What length crosses half of the total assembled bases?
- Q — Quality beyond contiguity: What evidence checks accuracy, completeness and structure?
Explained Practice
- N50 = 2 Mb. Is the assembly 50% complete? No.
- What does the 50 refer to? The halfway point of cumulative assembled sequence length.
- Can an incorrect join increase N50? Yes.
- Can an assembly with lower N50 be more complete? Yes.
- Should scaffold N50 be compared directly with contig N50? Not as if they are the same metric.
- What context should accompany N50? Total assembly size, piece count, accuracy, completeness and structural validation.
- What one sentence protects you from the main error? “N50 describes contiguity, not percentage correctness.”
For Parents and Tutors: Build N50 With Paper Strips
Cut five paper strips of different lengths and write their lengths on them. Let the learner sort them longest to shortest. State the total length, calculate half, then add strips until the running total passes halfway. The length of the strip that crosses the line becomes the classroom N50.
Next, secretly tape two strips together in the wrong order. The N50 may improve even though the representation is now less correct. This physical trick teaches the key distinction better than a definition alone: longer pieces and correct pieces are related goals, but they are not the same measurement.
Authoritative Sources
- A Field Guide to Whole-Genome Sequencing, Assembly and Annotation: defines N50 as a contiguity statistic and explicitly notes that N50 does not contain information about assembly accuracy.
- NCBI Genome Assembly resources: reference material for interpreting genome assembly records and their statistics.
- NHGRI — Human Genome Reference Sequence: useful background on what assembled reference sequences are used for.
- SEAB — 2026 PSLE Science syllabus: assessment objectives include interpreting and analysing information and evaluating observations, information and methods.
The Quiet Return
When a report gives one impressive number, ask what that number was built to measure. For N50, the answer is contiguity. Let it tell you about the pieces. Make accuracy, completeness and correctness show their own evidence.
