In the context of the metagenomics work: I've noticed that, in the case where no reads are aligned to a contig, samtools view --> samtools bam2fq will produce an empty FASTQ file containing zero reads. When such an empty file is provided to LJA, the full LJA pipeline is still run, producing a directory containing empty outputs (e.g. a GFA file with zero segments / links, an empty assembly.fasta file).
This is mostly problematic on the part of the user :), but -- if it's feasible to implement -- I suggest that these zero-read inputs should cause LJA to fail early with an error message. This would alert the user that their inputs are malformed.
In the context of the metagenomics work: I've noticed that, in the case where no reads are aligned to a contig,
samtools view
-->samtools bam2fq
will produce an empty FASTQ file containing zero reads. When such an empty file is provided to LJA, the full LJA pipeline is still run, producing a directory containing empty outputs (e.g. a GFA file with zero segments / links, an emptyassembly.fasta
file).This is mostly problematic on the part of the user :), but -- if it's feasible to implement -- I suggest that these zero-read inputs should cause LJA to fail early with an error message. This would alert the user that their inputs are malformed.