Plant stress speaks in four dialects. One bench reads them all.

• Pl@ntNet names the species and screens known diseases. Each result carries its own confidence.
• A segmentation model outlines symptomatic regions and measures their area: chlorosis, necrosis, leaf spot, rust, fungal, pest damage, other. A healthy region is not a symptom.
• A documented rule grades healthy, mild, moderate or severe from symptomatic area, symptom severity and disease confidence. Thirty percent symptomatic area saturates the area term. The grade quotes the numbers that produced it.
• Symptoms map to a literature metabolite table, which is not a measurement from the photo. A rules engine drafts a treatment plan you can keep or edit: Foliar Spray, Nutrient Supplement, Irrigation Adjustment, Environmental Adjustment, Pest Control, Monitoring.

• Species and confidence
• Disease findings
• Symptom overlays
• Stress grade
• Metabolite map
• Treatment plan

Example readout: mild stress, 11.8% symptomatic area.

If segmentation is not configured, the bench says “Segmentation not configured” and draws no overlays. Species, diseases and disease-driven stress still render.

• Computes a real group comparison, log2 fold change and significance, only when the table has the groups to support it.
• Draws the volcano so up- and down-regulated genes separate.
• Annotates with Gene Ontology. The interpretation reads the pattern and respects direction of regulation. It does not treat every change as activation.
• A bare gene list gets classification and GO context, never fabricated statistics.

• log2FC table
• Volcano plot
• GO terms
• Interpretation

Significant up = clay · significant down = chloro · noise = faint ink.

• Matches compounds to a curated catalog, including aliases.
• Grades each marker against its published literature band: below, in band or above.
• Combines graded markers into a stress level, a likely stress type and a duration band: hours, days, chronic, or no clear duration signal.
• Keeps the result paginated, searchable and sortable down to the marker.

• Per-marker grade
• Stress level
• Stress type
• Duration band

Example: proline 4.2× above band, osmotic stress.

When peak area is not µmol/g, the result is a within-file or literature-range estimate, not a lab diagnosis.

• The reference is one of eleven: Arabidopsis thaliana (thale cress), Oryza sativa (rice), Zea mays (maize), Solanum lycopersicum (tomato), Solanum tuberosum (potato), Glycine max (soybean), Triticum aestivum (wheat), Populus trichocarpa (poplar), Medicago truncatula (barrel medic), Nicotiana tabacum (tobacco), Chlamydomonas reinhardtii (green alga).
• Genes are located and annotated against that reference. Defense and stress roles are flagged only from real Gene Ontology terms, with evidence strength. A gene id alone never asserts a stress role.
• An optional stress filter. Abiotic: drought, heat, cold, salinity, flooding, oxidative. Biotic: pathogens, fungi, bacteria, insects, diseases.
• The result is a report you can file and reopen. An unknown organism, or an analysis with no species, is stated. It is not guessed.

• Gene annotations
• Stress-role flags
• GO evidence
• Full report

Example flag: GO:0006952 · defense response.