Peer-reviewed Publications and Independent Review of Algorithmic Performance
As developers of scientific software and algorithms we of course have our own approaches to validating the performance of our software tools. We do spend a lot of time testing, validating and improving our algorithms but, clearly, we apply our tools in a manner that might be different from the users of our software. This is because of our deeper understanding of our software, option settings etc. Despite our best efforts to document and improve ease of use different performance in the hands of different users is, in some ways, inevitable. Therefore, we see the results of our users as more of a test of the performance of our software and, where appropriate, we coach and mentor our users as to how to get improved performance from our tools. One of the common issues is that users treat all docking software in the same way. Therefore, they assume that because they have to do a lot of preparation of input data for software applications such as GLIDE that they have to do the same preparation work for eHiTS. This is not true. In fact, preparing data in the same way for different software tools can be potentially disastrous and comparing the results is invalid.
With this in mind we really appreciate the efforts of independent researchers to perform comparisons of performance of different software tools AND put in the work to understand, in depth, how to use the software tools properly. This includes data preparation, setting the appropriate parameters and examining the output in detail. Our experience is that eHiTS outperforms most other docking software available in the marketplace. This has been validated a number of times by our users in their feedback to us, but unfortunately not always published for public consumption. An example of such performance was shown in our examination of a Merck dataset as discussed here:
http://www.simbiosys.ca/ehits/ehits_enrichment.html and shown below in the image. We’ve seen such performance capabilities many times.
In the past few weeks we have received a number of manuscripts from our users. These have been submitted for peer-reviewed publication and represent the application of eHiTS to their problems and, in some cases, to the comparison of performance of eHiTS relative to other software packages. What we continue to see is a consistently enhanced performance of eHiTS over other software packages with eHiTS outperforming other algorithms, both commercial and in-house developments. We are, of course, proud of such achievements and feel validated in our own testing when our users can obtain such results independent of our coaching. These publications should be released in the near future and we will of course point you to them as they are released.