Consensus bayesian trees of RhcI T3SS cluster gene families

A custom set of 155 genomes was gathered covering 28 different Bradyrhizobium species, as well as selected genomes from the genera Nitrobacter and Rhodopseudomonas as outgroups. Genomes listed in Table S1 were downloaded from NCBI RefSeq or GenBank databases and used as input for the bioinformatic pipeline Pantagruel to build a phylogenomic database. In short, coding sequences (CDS) and the corresponding protein sequences were extracted from the RefSeq or GenBank annotation and then clustered into homologous gene families using MMSeqs2. Homologues were aligned with ClustalOmega and reverse-translated into CDS alignment with PAL2NAL. <br>Bayesian samples of gene trees were inferred from CDS alignments with MrBayes (v 3.2.6) using a Metropolis-coupled Monte-Carlo Markov chain (MCMCMC) approach, with two independent sets of four chains (one cold, three heated) running for 2,000,000 generations and sampled every 500; presented trees are majority-rule consensus trees, discarding the first 25% of the chain as burn-in.<br>Consensus gene trees are presented for gene families with homologs (based on the best BlastP search hit) to the reference CDS sequences in RhcI cluster from <i>Bradyrhizobium japonicum</i> reference strain USDA110, as indicated by the bll/blr locus tag and the corresponding NCBI RefSeq protein identifier.<br><br>