Null results for statistics parameters
CEPHAS-01 opened this issue · comments
Hi,
I just ran gfastats on a gfa file produced by hifiasm in trio mode to obtain the summary assembly statistics on scaffold length, N and L statistics etc. The command is as follows:
gfastats -f $inFile -t --stats > $outFile
The output I got is 0 for most of the parameters reported.
scaffolds 0
Total scaffold length 0
Average scaffold length nan
Scaffold N50 0
Scaffold auN 0.00
Scaffold L50 0
Largest scaffold 0
Smallest scaffold 0
contigs 0
Total contig length 0
Average contig length nan
Contig N50 0
Contig auN 0.00
Contig L50 0
Largest contig 0
Smallest contig 0
gaps in scaffolds 0
Total gap length in scaffolds 0
Average gap length in scaffolds 0.00
Gap N50 in scaffolds 0
Gap auN in scaffolds 0.00
Gap L50 in scaffolds 0
Largest gap in scaffolds 0
Smallest gap in scaffolds 0
Base composition (A:C:G:T) 0:0:0:0
GC content % nan
soft-masked bases 0
segments 21136
Total segment length 6154411558
Average segment length 291181.47
gaps 0
paths 0
edges 59590
Average degree 2.82
connected components 126
Largest connected component length 1596008570
dead ends 716
disconnected components 167
Total length disconnected components 115218292
separated components 293
bubbles 1052
circular segments 2
I am using the latest release (v1.3.6) of gfastats, extracted and compiled from the "gfastats.v1.3.6.tar.gz " file.
Am I running the command correctly?
Hi @CEPHAS-01,
Thanks for reaching out. Sorry I realize that this is a bit confusing. Conceptually, we do not consider a 'contig' as defined in the GFA unless there is an actual path (potentially involving multiple segments) that define its sequence. This is an attempt to distinguish contigs from segments.
In practice, just add the option --discover-paths and it will generate a path for each segment, thus generating contigs that can then be evaluated in the stats.
I'll add a comment in the readme as well.
Hi Giulio,
Thanks for the prompt response.
It sure works well now with the --discover-paths option.
Would be nice to have that as a comment in the README as you suggested, so that other users would be rightly guided.
done, thank you for the input!