Fig. 1

Orthograph workflow. From a set of reference proteins (1), the proteins are clustered to form orthologous groups (OGs) (2). These OGs are aligned to construct profile hidden Markov models (pHMMs) (3). The pHMMs are used to search for candidate orthologs in the target library (4). Each of the obtained hit amino acid sequences (5) is used as a query for a BLAST search in a database comprising all reference proteins (including the ones forming OGs) (6). Search results from both forward and reverse searches (7) are collated and sorted by bit score, with the reverse search result order being subordinated to the forward result order (8). This list is evaluated in descending order: if the reverse search hit a protein that is part of the OG used for the forward search, the candidate ortholog is mapped to the OG (9)