FASTA File Statistics: Sequence Count, Length Distribution, and N50

When you receive an assembly result or a batch of sequences, the first thing to check is the basic summary statistics.

N50 is the most commonly misunderstood metric. The definition: sort all contigs by length from longest to shortest, accumulate lengths in order, and when the cumulative length first reaches half the total assembly length, the length of that contig is the N50. It measures assembly contiguity — a larger N50 means longer fragments account for a greater proportion of the assembly.

L50 is the number of contigs needed to reach half the total length. N50 and L50 are a pair, and they move in opposite directions.

Note that N50 has a well-known flaw: it is sensitive to total assembly length. If an assembly is missing a large number of short contigs, the total length shrinks and N50 actually looks better. N50 must therefore always be read alongside total length and sequence count — a single number in isolation is meaningless.

All calculations run entirely in the browser; sequences are never uploaded. For large files, work with a subset first.

How N50 is calculated: a small example

Five contigs with lengths 100, 80, 60, 40, 20; total length 300, half-length 150:

Cumulative 100          < 150
Cumulative 100+80 = 180 ≥ 150  ← first to reach half
N50 = 80 (length of that contig)
L50 = 2 (took 2 contigs)

N50 can be inflated — this is its biggest problem

Suppose an assembly consists of 100, 50 plus ten contigs of length 10:

Nothing has improved — simply discarding short fragments doubles N50. Because the denominator of N50 is the assembly total length, when the total length shrinks the half-length threshold drops with it.

NG50 was designed to fix this

NG50 replaces the denominator with the estimated genome size rather than the assembly total length. Using the example above with a genome size of 250:

With a fixed denominator, discarding fragments no longer helps. When comparing different assemblies, as long as the genome size estimate is consistent, NG50 is far fairer than N50. This tool reports N50/L50 (no extra input needed); to compute NG50, substitute your genome size estimate manually.

contig N50 and scaffold N50 are not the same thing

Scaffolds are built from contigs joined by runs of Ns, and their lengths include unknown bases. Scaffold N50 is typically much larger than contig N50; comparing the two numbers directly is meaningless. When reading assembly quality reports in the literature, first confirm which one is being reported.