MAKER annotation

The final annotation of the genome was performed using MAKER2 v2.31.10 [102], which combines empirical evidence and ab initio gene prediction to produce final annotations. We used MAKER2 for three cycles of gene predictions. First, the Iso-Seq data were used as evidence for training MAKER2 for gene predictions. We also used transcriptome and peptide sequence data from An. gambiae (PEST4.12) and An. funestus (FUMOZ 3.1) as alternative evidence to support the predicted gene models. Prior to gene annotation, repeats were masked using RepeatMasker included in MAKER2. Mapping of EST and protein evidence to the genome by MAKER2 using BLASTn and BLASTx, respectively, yielded 12,324 genes transcribing 14,888 mRNAs. The output of first round gene models were used for the second round, where MAKER2 ran SNAP and AUGUSTUS for ab initio gene predictions. Next, another round of SNAP and AUGUSTUS predictions were performed to synthesize the final annotations that produced 14,966 genes, transcribing 16,559 mRNAs. In total, we identified 56,388 exons, 9791 5′-end UTRs, 9290 3′end UTRs, and 503 tRNAs (Table ​(Table1;1; see the “Materials and methods” section). We also predicted ab initio an additional 14,192 mRNAs/proteins but due to weak support they were not considered. The final MAKER annotation was assessed using recommended AED and Pfam statistical metrics.

The gene models were functionally annotated in MAKER2 v2.31.10 through a homology BLAST search to UniProt-Sprot database, while the domains in the annotated proteins were assigned from the InterProScan database (Additional file 1: supplementary text). We compared our gene model annotations with the draft assembly using OrthoFinder v2.3.7 [103] The GO enrichment analysis was performed in PANTHER v15.0 using PANTHER GO-SLIM Biological Process annotation data set [104]. Further, the orthologous top 1% of An. stephensi upregulated gene protein sequences in An. gambiae were identified by OrthoFinder v2.3.7.