On 2017 Feb 20, Stuart MacGregor commented:

Thank you for the bug report Christoph. Following your bug report (January 2016), the bug was fixed on the VEGAS2 web-based version in January 2016 but unfortunately not until January 2017 on the VEGAS2 offline version. Users who ran the top % test (not the default option) using VEGAS2 prior to these dates (or using the now retired VEGAS1) should re-run their analysis using VEGAS2. There is an FAQ item on this at https://vegas2.qimrberghofer.edu.au

On 2017 Jan 18, CHRISTOPH LANGE commented:

The VEGAS/VEGAS2 software can provide p-values for the top-percentage statistic that are anti-conservative, as the computation of the null-distribution is incorrect. In our technical report, we discuss the issue and provide code that, if included in VEGAS/VEGAS2, will provide correct p-values.

http://biorxiv.org/content/early/2017/01/17/101014

On 2017 Jan 18, CHRISTOPH LANGE commented:

The VEGAS/VEGAS2 software can provide p-values for the top-percentage statistic that are anti-conservative, as the computation of the null-distribution is incorrect. In our technical report, we discuss the issue and provide code that, if included in VEGAS/VEGAS2, will provide correct p-values.

http://biorxiv.org/content/early/2017/01/17/101014

On 2017 Feb 20, Stuart MacGregor commented:

Thank you for the bug report Christoph. Following your bug report (January 2016), the bug was fixed on the VEGAS2 web-based version in January 2016 but unfortunately not until January 2017 on the VEGAS2 offline version. Users who ran the top % test (not the default option) using VEGAS2 prior to these dates (or using the now retired VEGAS1) should re-run their analysis using VEGAS2. There is an FAQ item on this at https://vegas2.qimrberghofer.edu.au