Controls the maximum number of sequences included in the analysis.

If your alignment contains more species than this cap, the pipeline preserves the selected reference sequence and prunes the tree/alignment down to this threshold based on the taxa order.

Key Impacts:

• Selection Mechanism: The user-selected reference sequence is moved to the absolute front of the list to guarantee its inclusion. To maximize the phylogenetic diversity and preserve the overall tree span/length, the remaining sequences are selected using an exact greedy Phylogenetic Diversity (Faith's PD) maximization algorithm. This deterministically selects species that add the maximum evolutionary branch distance to the existing subtree, capturing the deepest splits and broadest taxonomic span. If no tree is provided, it falls back to systematic even-spacing sampling along the alignment.
• Computational Complexity: Reduces the time needed for client-side calculations, such as the O(N^3) Classical MDS coordinate extraction and HyPhy branch estimation. A hard upper limit of 512 is enforced to prevent the web browser from running Out of Memory (OOM) or crashing.
• Phylogenetic Resolution: Dropping species can prune away key intermediate nodes or sister lineages, potentially reducing the model's statistical sensitivity to detect episodic diversifying selection at certain sites.
• Inference Fidelity: The surrogate model's learned phylogenetic bias is optimized for balanced input taxonomic density. A default cap of 256 offers a robust trade-off between client-side latency and evolutionary signal depth.