Software Debugging is a tedious and costly task in software development life-cycle. Thus, various automated fault localization approaches have been proposed to address this problem, among which, spectrum-based fault localization has attracted a lot of attention. Using various formulas, known as ranking metrics, spectrum-based fault localization techniques assign scores to the entities of programs (e.g., statements) based on their suspiciousness of being the root cause of failures. Despite the obvious advantages of spectrum-based fault localization techniques, such as being lightweight, they cannot effectively locate faults in every program owing to the fact that they do not consider the characteristics of the programs. We believe that program characteristics can be helpful at finding the right ranking metrics for programs, and they can assist at combining several existing ones to produce a customized ranking metric specific to a given program.In this paper, we have proposed an approach which combines 40 different ranking metrics to generate a new ranking metric specific to a given program. Employing mutation testing operators, the proposed approach retrieves information from the program and then, using different preferential voting systems, it combines various ranking metrics based on the collected information. We have evaluated our approach on 154 faulty versions from eight different programs of Space and Siemens test suite and compare it with nine state-of-the-art ranking metrics. The experimental results indicate that the ranking metrics generated by our approach is superior with respect to evaluation metrics such as the Exam score and TOP-N.