Trouble Shooting

The X-Score program is a simple and user-friendly program. If you have experienced any problem when using this program, you may contact the author at the following address. We will try our best to get back to you. However, before you contact us, please make sure you have gone through this manual and it does not have the answer to your question. For general questions about how to run a program on Unix/Linux platform, please consult with a computer expert instead of us.

Renxiao Wang, Ph.D.
Research Investigator
Department of Internal Medicine, Hematology/Oncology Division
University of Michigan Medical School
Medical Science Building I, Room 2423
1150 W. Medical Center Drive, Ann Arbor, MI 48109, U.S.A.
Tel: (734)764-2449 Fax: (734)764-2532

We are in the process of compiling a more complete FAQs for X-Score. Any suggestion or comment on the program is highly appreciated. We have benefited so much by communicating with the X-Score users.

FAQs of X-Score

Questions related to the preparation of input files

Q: Why does the program prompt "cannot find binding pocket residues" and then stop?

A:  This happens when the ligand is apart from the target protein. Please remember X-Score will not do docking for you. Before applying X-Score, please make sure that the ligand has been docked into the binding site of the target protein and, more importantly, have been saved in the same coordinate system. The simplest way for checking this is to load the ligand and the protein on the screen to see if they indeed form the complex in your desired way.

Q. There are some water molecules in my protein PDB file. Is X-Score able to consider them?

A: The answer is no. X-Score will simply neglect all water molecules. In fact, almost all of today's scoring functions do not have a good idea for handling water molecules even when they exist on the protein-ligand binding interface. This is something we are working very hard on and we expect the next version of X-Score will be able to handle this.

Q. My target molecule is a DNA molecule. Can I apply X-Score?

A: No. X-Score is an empirical scoring function developed for evaluating protein-ligand interactions. Its application should not be extended to DNA-ligand complexes.

Q. I noticed that X-Score requires polar hydrogen atoms on the protein and all hydrogen atoms on the ligand because X-Score uses a hydrogen-bonding term. Do I need to optimize the hydrogen atom positions while keeping the heavy atoms fixed? 

A: It is not necessary. X-Score needs hydrogen atoms only for atom typing. The coordinates of all of the hydrogen atoms, on both protein and ligand, are not used at all in computation. Therefore, the coordinates of hydrogen atoms do not need optimization. By the way, if you want to add all hydrogen atoms (polar and non-polar) to the protein, it will not hurt.

Questions related to the scoring functions

Q: How to choose among the three scoring functions available in X-Score?

A:  There are three empirical scoring functions implemented in X-Score, namely HPScore, HMScore, and HSScore. They only differ in the algorithm for calculating the hydrophobic effect term. Statistically, the accuracy of these scoring functions is comparable to each other, at least for the protein-ligand complex set we have used for developing X-Score. By default, all the three scoring functions are enabled and the final result is the average of three predictions. This practice is consistent with the consensus scoring strategy and is generally acceptable.

However, it is also possible that, for the protein you are studying, one particular scoring function performs better than the other two. In such a situation, using that scoring function alone may lead to even better results. You may want to perform a test run first to figure out which scoring function performs better for your target.

Started from version 1.2, users are also allowed to set the coefficients in each scoring function. You may want to utilize this option to emphasize or underscore certain types of interactions between protein and ligand to achieve better structure-affinity correlation.

Q. Why X-Score's result is always a small number while some other scoring functions give scores around several hundred?

A: Because the scores are given in different units. X-Score gives dissociate constant of the given protein-ligand complex in negative logarithm units (-logKd). For example, for mili-molar (mM) affinity, -logKd = 3.0; for micro-molar (uM) affinity , -logKd = 6.0; while for nano-molar (nM) affinity, -logKd = 9.0. So, X-Score's result is usually a positive number under 10.

Q. Why in the output table some of the ligand molecules show score of zero?

A: Because you have let the program to use chemical rules to pre-screen all of the ligand molecules. If any molecule does not meet the chemical rules you set in the input file, it will be skipped in the later scoring process and therefore get a score of zero.

Q. Why cannot I accurately reproduce the original results reported in your JCAMD paper for the dataset included in the X-Score package?

A: That JCAMD paper reports our early work on the development of X-Score, which was mostly done in early 2000. Since then, we have made many modifications to the algorithm and also the program. Particularly, X-Score version 1.2 has been re-calibrated with a new training set. Thus, you should not expect the current version of X-Score to reproduce those original results by 100%. But one thing is sure: all of these modifications only lead to better performance.

Questions related to the chemical rules

Q. Why to use chemical rules in pre-screening?

A: It is well-known that drug-like molecules usually have a specific range of physicochemical properties.  At present, this feature is typically described by some knowledge-based rules, such as the famous Lipinski's "rule of five". Our own experience in applying X-Score in virtual library screening also suggests that by turning on the chemical rules for pre-screening, the false positives in the final results could be largely reduced. We recommend you to apply these chemical rules in structure-based virtual screening unless you have better ideas of how to select your compounds.

[Content] [Introduction] [Download] [Installation] [Usage] [Utilities] [Trouble Shooting] [References]