genomicintelligence.ai

Genomic Intelligence DNA analysis over MCP. Six inference tasks (promoter, splice, enhancer, chromatin, expression, annotation), plus a composite annotation→expression workflow. Research and development use. Not for clinical or diagnostic decisions. For high-volume or latency-sensitive use, contact contact@genomicintelligence.ai to provision a production-shaped deployment. Recommended flow to avoid bloating context with large sequences: 1. Acquire a sequence as a *handle* (fetch_ensembl_sequence, fetch_gene_for_expression, load_local_fasta, or store_inline_sequence). 2. Pass the returned `sequence_ref` to a predict_* tool. Small sequences may instead be passed inline via `sequence`. Each task has its own length bounds (promoter 300, splice 100, enhancer 50, chromatin 200, gene finding 1,000, expression 9,198 bp minimum; 500,000 bp maximum for all of them) — see gi://docs/tasks. A minimum is admission control, not the model's regime: a shorter-than-context-window sequence is still scored, against a padded window, so compare your length with bio_spec.context_window_bp from gi://models before reading much into a short-input result. Reference context lives in resources: gi://models, gi://docs/tasks, gi://sequences, gi://account.