Abstract
This formal correction addresses critical updates to the 2012 study "Towards a Better Detection of Horizontally Transferred Genes by Combining Unusual Properties Effectively." The primary focus is on ensuring the reproducibility of the Hgtident tool by providing updated data access and clarifying the mathematical definitions of the evaluation metrics used to validate the method.
Introduction
The original research introduced Hgtident, a method designed to improve the detection of Horizontally Transferred Genes (HTGs) by integrating various genomic "unusual properties," such as GC content and codon usage bias. Following publication, several issues were identified regarding the availability of the validation database and the precision of statistical terminology used in the text. This correction ensures that the community can accurately interpret and replicate the study's findings.
Data Availability and Reproducibility
The original study relied heavily on the HGT-DB database for validation. However, this database has since been decommissioned and is no longer accessible at its original URL. Furthermore, the annotations within HGT-DB are now considered outdated by modern genomic standards.
To maintain the scientific integrity and reproducibility of the work, the authors have migrated the code, data, and implementation details to a public GitHub repository: https://github.com/gpanda1/for_hgt. Researchers are encouraged to use this repository for any future comparative studies, while remaining mindful that the "ground truth" reflects the state of genomic knowledge in 2012.
Clarification of Evaluation Metrics
A significant portion of this correction involves the "Evaluation Criteria" section. While the original paper's calculations were correct, the textual definitions of Type I and Type II errors were mislabeled. The corrected definitions are as follows:
- Type I Error (False Positive Rate): Defined as the ratio of false positives to the total number of actual negatives: $$\text{Type I error} = \frac{FP}{TN + FP}$$
- Type II Error (False Negative Rate): Defined as the ratio of false negatives to the total number of actual positives: $$\text{Type II error} = \frac{FN}{TP + FN}$$
The performance of Hgtident was evaluated using these metrics to derive the Mean Error and Recall:
- Recall (Detection Rate): $\frac{TP}{TP + FN} \times 100$
- Mean Error: $\frac{\text{Type I error} + \text{Type II error}}{2} \times 100$
Conclusion
The authors emphasize that these terminological corrections do not alter the original findings or the reported performance of the Hgtident tool. The underlying logic used in the software implementation was consistent with the corrected formulas. By providing these clarifications and a new data repository, this correction ensures the continued utility of the study in the field of horizontal gene transfer detection.
Summary updated