Different assembler and difference in genomic variants. Is it possible?

Question

[Rashedul Islam]

Hi,

I assembled a viral genome using two different assemblers e.g., Megahit and Metaspades. Both the assemblers uses de bruijn graphs but might use different k-mers because both the assemblers uses multi k-mer approach.

I aligned the contigs to the reference and called variants using BCFtools. I am getting 18 SNPs that are found in one of the assemblers. Im surprised to see the degree of disagreement in variant calling between two assembly methods. Is it expected or what would be the possible explanation for that?

[swbarnes2]

How likely is it that your sample is not clonal?

[Rashedul Islam]

Its a ssRNA virus but not sure about clonality. Its almost impossible to have two different strains in the sample.

[andres.firrincieli]

As fars as I know metaspades perform a read correction error step before starting the assembly step

[Rashedul Islam]

Yes, thats a great point. Thanks!