Why does FragGeneScan, a metagenome gene prediction tool, use a 6-periodic HMM?

Question

[mimi]

Hello, I would like to know why the metagenome gene prediction tool, FragGeneScan uses a 6-periodic HMM to represent the coding region.

Why did this tool choose a 6-cycle HMM instead of a 3-cycle HMM even though the codon is composed of 3 nucleotides?

I couldn't find this answer as far as I read the paper.

https://academic.oup.com/nar/article/38/20/e191/1317565

six-periodic hidden Markov model (HMM). (The figure below (i) and (vii) [Rho M, et al. 2010])

Thank you!

[5heikki]

Because it assumes that the input sequence is from double-stranded DNA? The same reason why e.g. blastx searches in six reading frames..

[mimi]

Thank you, 5heikki. In my view, it doesn't seem to represent a difference in reading frames because this tool models forward direction (figure (i)-(iii)) and reverse direction (Figure (v)-(vi)) separately and represents each in 6 states (diamond in the figure). Please tell me If I understand your answer wrong.