I'm new to spades and i've started on some genomes which i previously assembled in velvet (~6 Mb chromosome plus one ~40 Kb plasmid, and a few long repeats ~5Kb) it's old illumina data with 100 bp reads and 100 to 400 bp paired end inserts (kinda poor quality paired ends, with a bimodal length distribution in some samples unfortunately).
I've tried both the standard whole genome version and the plasmid spades version, treating the data as paired end, or as just single reads.
I'm not very impressed with the results.
The first problem is i'm getting a lot of junk contigs, probably due to bubbles in the graph, but these are pretty easy to exclude with a coverage cut off (is there a quick way to do that automatically?).
but, the bigger problem is that it seems to be joining a lot of things which shouldn't be joined.
Is there a setting somewhere to reduce the risk of misassemblies? i don't care if the contigs are smaller, but it's important that they are right. I want real quality not just an impressive N50.
I've tried both the standard whole genome version and the plasmid spades version, treating the data as paired end, or as just single reads.
I'm not very impressed with the results.
The first problem is i'm getting a lot of junk contigs, probably due to bubbles in the graph, but these are pretty easy to exclude with a coverage cut off (is there a quick way to do that automatically?).
but, the bigger problem is that it seems to be joining a lot of things which shouldn't be joined.
Is there a setting somewhere to reduce the risk of misassemblies? i don't care if the contigs are smaller, but it's important that they are right. I want real quality not just an impressive N50.