Pipeline overview of AGAPE for yeast.

<p>The pipeline consists of three parts; (a) assembly, (b) annotation, and (c) variation. Cylinder shapes indicate data, shaded cylinder final result data, arrows data flows, rectangular shapes programs, and dotted rectangular external package tools that are not included in our pipeline. After all ambiguous and low quality reads are discarded, the remaining reads are processed to generate assembly contigs (a). The assembly contigs from (a) are used as the input to annotate their genomic features including both reference ORFs inferred by a homology-based method and non-reference ORFs predicted by <i>ab initio</i> methods (b). Fungal (including yeast) protein and EST databases are used to accurately predict annotations. In a post-process annotation step, annotated ORFs are refined and corrected as shown in (b). For variation detections, the reads remaining after the error-correction step are mapped to the reference genome in (c). The procedure (c) then forks into two branches; one for unmapped and another for mapped reads. The unmapped reads are assembled in the manner described in (a) to contigs, then compared with the assembly contigs from (a) and annotation results from (b) to identify newly inserted sequences and ORFs that are not present in the reference genome. For the mapped reads, the mapping information is used for the HugeSeq pipeline that detects variations including SNPs relative to the reference. The SNP calls and the non-reference features identified in (c) can be used for further variation analysis using external tools, e.g. the Galaxy genome diversity tool and various R packages.