**5. Conclusions**

In conclusion, when estimating kinetic parameter values, PSOHS performed better than the downhill simplex method and SA, as shown by the smaller standard deviation in PSOHS. Moreover, PSOHS is less time-consuming. The lower nonlinear least squared error of PSOHS also proves that this algorithm is more accurate compared to SA and the simplex downhill method. Future lines of research are going to focus on including different performance measurements and algorithms and on comparing how they affect the performance of PSOHS. Only one dataset has been used in this research due to an unavoidable constraint. However, more datasets can be used in future work. Large-scale metabolic parameter estimation is preferable. However, the inclusion of more datasets poses a bigger challenge in that the parameters of every single gene and its product will be needed in order to be estimated. It leads to large-scale metabolic parameter estimations [20]. PSOHS should be further expanded to that scale with the aim of resolving the problem. Besides, the shortcomings of the existing HS, such as the fine-tuning ability of the algorithm, can also be improved in future work [21].

**Author Contributions:** Conceptualization, M.S.R. and M.S.M.; methodology, M.S.R. and M.S.M.; project administration, M.S.M.; resources, M.S.R.; software, M.S.R.; supervision, M.S.M.; writing—original draft, M.S.R.; writing—review and editing, M.S.M., Y.W.C., Z.I., A.G.-B., P.C. and J.M.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** We would like to thank the Skim Geran Penyelidikan Fundamental (FRGS-MRSA) (no grant: R/FRGS/A0800/01655A/003/2020/00720) from Ministry of Education Malaysia for their support in order to make this research a success.

**Conflicts of Interest:** The authors declare no conflicts of interest.
