I want to try out IDBA-UD (idba v 1.1.1) as a metagenome-assembler, but am a little confused.
It seems idba_ud requires a single interleaved paired-end fasta-file as input.
However, i like to quality trim/quality-filter my data (e.g. with TRIMMOMATIC), and that means that some of the paired reads lose their respective "partner" (thus becoming unpaired reads). I do not like to completely drop these reads, because that means discarding potentially valuable sequence information.
With other assemblers, There is usually the option to include paired and unpaired reads in the same assembly. Is there a way to do this with idba_ud, also?
I really do not think the "--read_level_<2-5>" args are meant for this (they are meant for scaffolding and strictly require paired end info).
I am a little concerned about trusting the assembly on untrimmed data, because, as idba explicitely requires you to convert your fastq-input to fasta, idba cannot have any chance to do error correction based on quality values, am i right?
Does anybody have any experience on this (or can point me to a good idba-tutorial as I can't seem to find any helpful documentation for this tool)?