Hi @Jawairia,
Please see this post. It sounds like you have the same issue as the others in that thread.
As a workaround, I recommend A) removing that sequence, B) increase the --p-identity parameter (e.g., to 0.9) for extract-reads, or C) do not do the read extraction (this boosts accuracy by a bit compared to full-length, but it is not a game-changer).
I hope that helps!