Which method answers which question
A short guide to the methods, with links to their full descriptions.
Reference
The Data Analysis menu offers six methods and the Data Download menu two export jobs. Each answers a different question about a set of sequences. This page says what each one computes as implemented today and when to use it; the full description, parameters, output files and limits are on each method's own page.
How to read this page
Every method takes the records you ticked, keeps those with the chosen marker (where it filters by marker) and a sequence inside the length range, drops any record whose sequence is more than 10% ambiguous bases, and then runs. Methods that compare sequences position by position need an alignment: choose MAFFT under Align Sequences unless your sequences are already aligned. Some form options are accepted but not applied; the most important are noted below and Known limitations of the methods lists them all.
Sequence Identification
Which reference records do these sequences most closely match?
The job runs a tiered BLAST search of each selected sequence against a reference database built from the platform's own records, all markers included. For each query it returns matching reference records ranked by identity, overlap and bit score, with their taxonomy, and a simple label: a near-exact or strong single-taxon match, a single-taxon match below 98%, an ambiguous top cluster when the best hits belong to more than one taxon, or no confident hit. Use it to see what a sequence resembles in the reference set. It assigns no rank and no confidence score, a sequence already in the database matches itself, and the library and Marker choices do not restrict the search. See Identification.
Distance Summary
How divergent are the sequences within each species, genus, subfamily, family and order?
The job computes a distance for every pair of sequences and assigns each pair to the lowest rank at which the two records share a name, then reports pair counts and the minimum, mean and maximum distance for each taxon at each rank. Use it to see how much variation sits inside species and between related species. The distance is always uncorrected p-distance whichever substitution model you pick; choose MAFFT and pairwise deletion, as complete fails. See Distance summary.
Barcode Gap
Does each species' largest within-species distance stay below its distance to the nearest other species?
For each species the job compares the largest distance among its own sequences with the distance to its nearest neighbour, looked for first within the same genus, and reports the gap. Species with no positive gap, and species represented by a single record (which have no within-species distance), are listed as failing. Use it to judge whether the species in a reference set are separable by distance. The job uses K2P whatever model you pick, a fixed length window of 200–700 bases rather than the slider, and pairwise deletion. See Barcode gap.
Diagnostic Characters
Which alignment positions distinguish a species from the other species of its genus in this set?
Within each genus, the job compares every species with at least the minimum number of records against its congeners and reports the columns where the species carries a state the others lack, classed by how cleanly the column separates them; paired columns are tried when no single column does. Use it when you want character evidence rather than a distance threshold. Positions are columns of this run's alignment, not gene coordinates, and results are relative only to the congeners present. Choose MAFFT; the grouping option is not applied, and the comparison is always species within genus. See Diagnostic characters.
Tree Construction
How do the selected sequences cluster?
The job builds a tree with the program you pick (QuickTree, FastTree, RAxML or IQ-TREE), labels the tips with the record fields you choose, and returns a multi-page PDF (100 tips per page) and a Newick file. Use it for a visual overview of a set. No rooting or outgroup step is applied, there is no bootstrap support (IQ-TREE adds SH-aLRT values), and the tree is always drawn in the same multi-page style coloured by phylum regardless of the style and colouring options. See Tree construction.
Sequence Composition
What are the base frequencies and GC content of each record and taxon?
The job reports A, C, G, T and GC percentages and sequence lengths per record, then summarises them by order, family and genus with a GC histogram. Use it as a quality check for unusual composition or to compare taxa. No alignment is needed. Missing taxonomy appears as a taxon named null. See Sequence composition.
Specimen Records and Sequences (Data Download)
Give me these records, or these sequences, as a file.
The specimen export writes the fields you tick to a TSV or CSV file; the Excel and JSON options fail, and only specimens that have a sequence are included. The sequence export writes a FASTA file with the header fields you tick; its alignment and length options are not applied. Neither produces a report. See Specimen and sequence exports and Downloading records, sequences and files.
Good to know
Limit: No accuracy, precision or validation figures exist for these methods. Treat the outputs as descriptive summaries of the records you selected, and check the notes above before citing an option.