N50 value of bacterial assembly is not half of total assembly size

Question

[analyst]

Dear scientists,

I have performed de novo assembly of bacterial WGS data using spades and abyss. However I am not sure which assembly approach is good to go because N50 value is not half or more than half of total assembly length.

All reads are QC passed (no adapter or primer sequences are found and base quality is above 30).

And also please suggest that how many contigs should I keep for further analysis. Length of reference genome is 5,240,075 bp.

Here i am attaching assembly reports please input your valuable suggestions and guidance.

Assembly with abyss

Assembly with spades

Assembly with spades (using --trusted-contigs flag where reference genome was used to guide assembly)

Assembly with unicycler

[GenoMax]

Assembly with spades (using --trusted-contigs flag where reference genome was used to guide assembly)

Looks like you pasted the stats for plain SPAdes assembly twice or vice versa. Since the last two stats are identical.

[analyst]

Thanks for informing GenoMax. I mistakenly pasted stats of spades (with --trusted-contigs option) twice. I have edited the post.

[Jeremy Leipzig]

N50 is the length of the size-ordered contig that puts you over that 50% threshold

http://jermdemo.blogspot.com/2008/11/calculating-n50-from-velvet-output.html

[analyst]

Thank you so much Jeremy for sharing useful link.

Answer 1

[Mensur Dlakic]

Your SPAdes assembly looks fine. N50 is not supposed to be half a genome. I would take a >5000 bp as a contig cutoff and that should be similar to your reference genome. This can be confirmed by mapping. You can lower the threshold maybe wo 2000 bp, but that will hardly be different from 5 kb.

[analyst]

Thank you @Mensur Dlakic for your valuable suggestion!

Please have a look on stats again I mistakenly pasted stats of spades assembly with --trusted-contig flag twice. Adding --trusted-contigs parameter brought difference in total contigs number and N50 value as well.

Please suggest which spades approach should I follow?

[Mensur Dlakic]

Not sure why you need me to answer this question, as the --trusted-contigs assembly is clearly the best. It has fewest contigs, largest N50 and the largest contig overall.

[analyst]

Thank you for your answer. Best,