After unsuccessfully aligning an extracted reverse complement 16S from an Illumina sequenced isolate, I found myself asking... how does the assembler actually know which way to print out the contig? Does it every happen that a metagenomic-assembled genome has a mix of plus and minus strands? Or is this a step most de novo assemblers take care of? How does one circularise such a genome if it is complete but with half the contigs in reverse?

At the graph building level, I guess entire contigs could be output in any direction. So how do draft genomes and bacterial genomes actually get refined to make sure there is one genome going in one direction?