Quartet Distance compares the relationships induced by each four-leaf subset of two trees. Read the unresolved-star convention, the denominator and the complete saved table before interpreting its figure.
91220f38… with the executed snapshot: twelve selected functions and three shared dependency files match. The full app files differ. This is bounded method evidence, not a new whole-app run or a biological benchmark.Product and app: Computational Biology · Tree Comparison & Topology Metrics · computational_.
1. Preserve the supplied trees and their order
The retained input names its first tree Binary AB / DE and its second Partial AB. Their Newick strings are:
((A,B),C,(D,E));
((A,B),C,D,E);
Both contain the same five leaf labels, A through E. The second leaves the relationship among C, D and E unresolved. No branch lengths are supplied. Keep both original strings, names and input order; the saved input file has its own byte hash, distinct from the application’s input fingerprint.
The source’s supplied-tree path requires at least four taxa and matching taxon sets across parsed trees. This route additionally checks unique, nonempty leaf labels. It compares the first two trees. Names such as NJ and UPGMA still occur internally; for supplied trees, those positions do not establish that either input was inferred by a particular algorithm.
The recorded native workflow supplied the named-tree bundle when the app opened, then used Run, the viewer and Export All. It did not exercise a Load dialog, launcher, hosted upload or sequence-inference path. Confirm that the available delivery path supplies the intended tree contract before planning a run. The retained checks are not general Newick-parser acceptance.
2. Select the quartet route explicitly
All twenty methods start selected in the reviewed source. For this example, retain only Quartet Distance, identified by quartet_. Check the selected-method record and completed-run status before opening its results. A completed route does not establish coverage of the other nineteen methods.
This calculation uses unrooted topology. It ignores branch lengths, including explicitly zero lengths, and does not contract an edge merely because its length is zero. The implementation copies unit edge lengths for its topology calculation; it does not replace the uploaded tree’s stored lengths. For a branch-weighted question, read the separate weighted path-distance guide.
3. Distinguish resolved and unresolved quartets
For each set of four taxa, the route recognizes three resolved pairings and an unresolved star. Two identical states agree. A star compared with a resolved pairing differs once, just as two different resolved pairings differ once. Unresolved here describes a supplied topology; it is not a missing numerical result.
The following five rows come from the retained independent reference, which projected structural edge splits onto each subset. This guide reads that reference without parsing the trees or calculating quartet distances again.
| Subset | First tree | Second tree | Same? |
|---|---|---|---|
ABCD | AB|CD | AB|CD | Yes |
ABCE | AB|CE | AB|CE | Yes |
ABDE | AB|DE | AB|DE | Yes |
ACDE | AC|DE | star | No |
BCDE | BC|DE | star | No |
Quartet distance counts differing four-leaf subsets. It is distinct from Robinson–Foulds distance, which counts differences between split sets in the described routes. Their units and denominators should not be interchanged. The TreeCmp paper’s quartet-metric section provides primary methodological context; it does not establish that this app executes TreeCmp or shares its performance.
4. Read all fourteen saved fields
The denominator is the number of four-taxon subsets, C(n, 4). The retained five-taxon example has five subsets, two differences and three agreements. The source puts numeric fractions in the table; its log displays those fractions to four decimal places. A rounded log line and the saved numeric field are different representations.
| Saved field | Retained value |
|---|---|
| QD | 2 |
| Total Quartets | 5 |
| Agreement | 0.6 |
| Normalized QD | 0.4 |
| Agreed Quartets | 3 |
| First Unresolved | 0 |
| Second Unresolved | 2 |
| First Resolutions | 5,0,0,0 |
| Second Resolutions | 3,0,0,2 |
| Resolution Order | ab|cd, a<b<c<d by leaf label |
| Scope | Unrooted topology; branch lengths ignored; star/resolved difference counts one |
| First Tree | Binary AB / DE |
| Second Tree | Partial AB |
| Ref | Estabrook et al., Syst Zool 1985; Sand et al., Bioinformatics 2014 |
Agreement 0.6 is the fraction of these five subsets with the same induced topology. It is not a probability that a tree is correct, a bootstrap support value or a confidence interval. A distance of zero can also occur when both inputs share an unresolved topology.
5. Match the figure to the table
The pie chart shows three agreeing and two differing subsets. The grouped bars show the four resolution-category counts for each tree: [5, 0, 0, 0] and [3, 0, 0, 2]. The labels a, b, c and d are sorted within each four-leaf subset. A category called ab|cd therefore describes a relative pairing, not one fixed pair of globally named taxa across every subset.
The recorded native viewer clipped long Scope, Resolution Order and Ref text. The complete CSV retains those fields. Preserve that visual limit and read the table outside the viewer; these browser-formatted tables do not establish that the native viewer has been repaired.
6. Keep the complete five-file export
The native viewer’s Export All action saved four listed artifacts and a separate manifest. The writing review checks every file identity and all 28 CSV data cells against the saved native table.
| Saved file | Contents and limit |
|---|---|
Quartet_ | Parameter/Value columns and all fourteen rows shown above, including the full scope and reference text. |
Quartet_ | The recorded agreement and resolution-category figure. Its file identity is retained without a new rendering. |
Input_ | A JSON record of the application, input fingerprint, selected route, selection notes and completed status. |
Summary.txt | The run summary. It does not replace the numerical CSV. |
artifact_ | Names, byte counts and SHA-256 hashes of the four artifacts. Keep the manifest itself too. |
This five-file set contains no full Newick input copy or per-subset resolution ledger. Keep the original named-tree input and the independent reference separately. File integrity does not establish scientific validity, and this bundle is not a demonstrated reloadable app session.
7. Preserve the source and evidence limits
The native pair used source f869b749…. The original guide review at 19:32 UTC on 11 November 2026 compared it with snapshot 25d5d907…. A separate saved-only review at 21:02 UTC that day compared snapshot 91220f38…: twelve selected input, route-selection, parser and quartet functions match the executed snapshot, as do three shared UI/export dependency files. The downloadable evidence preserves both dated records and identifies the residual probability-panel integration receipt. The full app files differ; that later panel correction adds no quartet run or export acceptance.
The retained candidate suite has 29 checks, including sixteen ordered pairs drawn from the same four small topologies. Those are not 29 independent datasets. Its earlier baseline has 21 failures among 27 checks, and five attempts stopped on shared-lock contention before execution; all remain preserved. The accepted native evidence is still one constructor-loaded five-taxon pair.
The source enumerates all C(n, 4) subsets. No large-tree performance, cancellation, real biological data, general parser, external executable comparison, hosted delivery or whole-product acceptance follows from this example.
Download exact source identities, saved fields, input and export records. No scientific computation, native Run, export, media capture or rendering was repeated for this guide. Continue with Results, Figures & Reporting or return to the written app guides.