The Easy Button and the independent-part controls call the same stage functions. This avoids maintaining a second scientific implementation for the automatic workflow.

The current Version 4.02 beta runs in the browser and does not require genome sequences to be sent to a Sequora server. Users should still follow their institution’s rules for handling sensitive or restricted data.

• The recommended baseline uses 1,000-base Query chunks and 10,000-base Reference chunks.
• Neighboring Reference chunks are concatenated into larger searchable records. This is different from describing the References as having a fixed 1,000-base overlap.
• For legacy 3-20 circular handling, Sequora builds an additional record from the final two Reference chunks followed by the first chunk.
• Version 4 also supports a boundary-window construction as an alternative circular mode.
• Reference records are accompanied by full IUPAC reverse complements.

• The current Version 4.02 beta collects distinct, non-overlapping Query k-mers starting at the first base of each Query chunk.
• It compares them with distinct, overlapping Reference k-mers beginning at every possible Reference position.
• The k-mer score is the percentage of distinct Query k-mers found in the Reference record.
• Repeated copies of the same k-mer are counted once in the current validated implementation.
• IUPAC symbols are treated as literal sequence symbols during k-mer matching and alignment; they are not expanded into sets of possible bases.
• Passing pairs use local affine-gap Smith–Waterman with the selected match, mismatch, gap-open, and gap-extension scores.

The Version 4 engine has been regression-tested against the validated Desktop Part 3 implementation for alignment results and k-mer filter decisions on the tested fixtures.

Reference labels are cleaned back to their original genome identity before grouping, while the full Query chunk label is retained. Every row tied at the maximum score is preserved so later stages can trace exactly which processed Reference records contributed.

Version 4 validates that Aligned is a non-negative integer, that a Query chunk count exists, and that Aligned cannot exceed Total Query Chunks. Invalid accounting stops the stage rather than producing a misleading value.

Many legacy distance columns use the form (1 − observed value ÷ largest observed value in that run) × 100. Because this normalization depends on the current collection, the same genome pair can receive a somewhat different normalized legacy distance when analyzed within a different group of genomes.

Version 4 places MARS Distance before Combined Metric 1 in metric discovery and reports.

These formulas describe how the current code transforms the measured alignment results. They do not convert sequence difference into elapsed time or identify a historical ancestor.

• When A→B and B→A are both available, the current matrix uses their average.
• When only one direction is available, the recommended setting uses that measured direction and counts the pair as single-direction. An advanced strict reciprocal option is also available.
• When neither direction is available, the selected missing-distance policy determines whether a recorded placeholder is inserted or the stage stops.
• Changing a display root in a future build will not change the distance matrix or the underlying unrooted topology.

Phylogenomics webs and complementary network language

The placeholder is computed independently for each metric as 1.10 × the largest real, finite symmetric distance observed for that metric. When no positive measured value exists, the present implementation uses a fallback value of 1.