AI-assisted gene comments - minor tweaks list

• Consider adding to the prompt guidance: "Preserve the kind of relationship described in the source. Do not reinterpret proximity on an expression map, pathway map, network, clustering diagram, or dimensionality-reduction plot as physical genomic proximity." after AGAP001212 + https://pmc.ncbi.nlm.nih.gov/articles/PMC3341590/ made an incorrect assumption about genomic colocalization: "Conserved active site residues suggest amidase activity typical of catalytic PGRP family members; genomic co-localization with antimicrobial effector genes further supports this functional assignment computationally." - Claude was fooled by this passage from the paper: "PGRPLB lies in a region populated by other effector genes and peptides (GAM1 (AGAP008645), LYSC7, DEF1 and cecropins 1-3),"

Review the cited PMC3341590 passage and the AGAP001212 example to understand the incorrect genomic-colocalization inference. Locate the AI prompt guidance in the repository, add the requested distinction between visual proximity and physical genomic proximity, and verify that the example no longer leads to that assumption.