# How each real value was produced (rerunnable)
- Source CDS: extracted from index.html RS.ref (2037bp) / RS.x3 (2025bp); verified byte-identical translation to NCBI NP_001011581.1 via efetch (NM_001011581.1_cds). GC 47.13%, 678/674aa.
- Homologs: NCBI efetch protein NP_001011581.1, ACS36224.1 (Bombus), AEQ57608.1 (Apis cerana partial 95aa, excluded from tree); UniProt REST A0A7M7GDM4 (Nasonia 777aa), P32023 (Drosophila for-PKG 934aa).
- Alignment/tree: pure-python Needleman-Wunsch (match+2/mismatch-1/gap-2) -> % identity matrix (data/identity_matrix.json) -> UPGMA teaching tree (data/tree.nwk). UPGRADED: MAFFT 7.526 alignment (974 cols, data/amfor_msa_MAFFT.fasta) + IQ-TREE 2.4.0 ML (ModelFinder BIC JTT+I+R2, 1000 UFBoot, Dmel outgroup) -> data/tree_ml_iqtree.contree (page figure). Bombus+Nasonia 90; X3 split 61 (weak, disagrees with UPGMA - uncertain). NW identity is tie-break-order-dependent (Nasonia-ref 74.4/74.6); ML uses the MSA directly.
- Domains: InterPro via UniProt A0ACM8PXT7 (exact 678aa NP_001011581.1 equivalent), data/interpro_Amfor.json: cNMP 102-217 + 220-335 (IPR000595), kinase 367-626 (IPR000719), AGC tail 627-678 (IPR000961), active site 491, ATP site 373-381. X3 coordinates mapped through the real ref-vs-X3 alignment. (Earlier PROSITE-style pattern scan kept in data/domain_motifs_pattern.json; it found Gly-rich 374-381 = annotated ATP site.)
- QTL: literature-curated (see CSVs). Amfor locus NC_037650.1:10187774..10107901 minus strand from NCBI Gene 406092 esummary. pln/sting QTL given as original LG/STS - lift to Amel_HAv3.1 with your QTL table.
- Expression: REAL counts - GEO GSE64315 (Khamis et al. 2015, PMID 26073445): CAGEscan brain tags, 8 nurse + 8 forager individuals; CPM normalize, Welch t on log2(CPM+1), BH FDR; 584 DE at FDR 5% (239 up foragers incl. Defensin/Mrjp6/Obp4). Amfor/per/Vg not DE at TSS level (consistent with alternative-TSS usage; total-mRNA literature says up). Full table data/de_table.csv, figure data in data/de_nurse_forager.json.
- Figures 13-16: gene model from GenBank NC_037650.1 slice 10107000..10189000 strand 2 (CDS-in-exons verified for all 4 isoforms; genomic = 10189000-slice+1); GC1/2/3 + GC skew in 60bp windows on real CDS; AA radar from real homolog composition; maturation timeline from dated literature findings (tooltips cite sources; timings typical, not exact). Earlier figures: identity heatmap (5x5 NW matrix), RSCU scatter (61 sense codons), PCoA (classical MDS on 1-identity, PC1 75.6%), sliding-window NG86 dN/dS (30-codon windows, mean omega 0.21 = purifying selection; lone capped spike at codon ~375 is a dS~0 artifact), motif alignment (all 5 motifs present in all 5 species), evidence network (literature-curated links, schematic layout). Remaining honestly-labeled limits: UPGMA kept for reference but ML shown (weak 61 node flagged); pattern motifs superseded by InterPro; literature QTL without genome coords (lift to Amel_HAv3.1); CAGE measures TSS tags (2 pooled-sequenced libraries of 8+8 barcoded bees), not full-length RNA-seq.