I have developed a software for de novo genome assembly. Its performance varies gradually according to how much data you employ. At initial stages it often produces contigs that look like that when aligned to a reference genome:
Initial-stage s-aligner assembly of SRR8357422 Genome fraction:98,8% - Largest Alignment: 61.544 - NGA50: 59.409 - LGA50: 2
If I add more data, the contigs end up being larger, but that obviously takes more time. My question is, how does this result compare to, for example, one like that?
SPAdes assembly of SRR8357422 Genome fraction:93,46% - Largest Alignment: 128.391 - NGA50: 128.391 - LGA50: 1
Note how you can construct a contig covering almost the full reference genome assembling green contigs in the first case. Therefore, a blast search of genes will likely let you find more genes given that it covers more genome fraction.
Apart from determining what option is more convenient, I would also like to know how much more convenient is to have both results. Or in which cases it's more convenient.
Last (but not least!), is there a way to generate automatically larger contigs from these overlapping contigs in assembly 1? I have tried using miniasm with no success. Even minimap2 seems not to find overlaps well enough. I am developing my own algorithm also for that but would like to know if there is another way.
[EDITED on 17/12 to add another example with real data for both assemblies]