[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [modeller_usage] validation



Adriana Esteves wrote:
I have read the answer you gave to simran jeet about residues in disallowed
position, sugesting  to  build multiple models and pick the best one. Your
answer conducts me to the following question. How to build multiple models,
with multiple templates or performing several models with the same template
and selecting that with the  lowest objetive value? In this case, how many
models are indicated to run?

Certainly if you have multiple possible templates, there may be some value in trying each of them (or a combination). But by "multiple models" we usually mean to build many models for the same template.

There is no rule on how many models to build, but if your models show a lot of variation in energy or structure, you probably need to build more to get adequate sampling.

A second doubt I have is about  validation. As the tutorial indictes,  one
should check the log file from the modeling run for runtime errors
(`model-single.log') and restraint violations. I did it but I do not know
how to interpretate the restrain violations. Can anyone help me to
interpretate them?

Generally speaking if a 'strong' restraint (such as distance) is violated by more than 2 or 3 standard deviations, it's worth investigating. Weaker restraints such as phi-psi are probably less of an issue.

I have also  run the file `evaluate_model.py' to
evaluate my model with the DOPE potential, but I do not know how to access
to gnuplot.

You don't have to use gnuplot. Any program which can plot the value in the rightmost column in the text file output by evaluate_model.py will do. For example, Excel works just fine too.

Finally I'd like to check the model with procheck_NT but
something is wrong with the pdb file.

Many validation tools are designed for X-ray structures, and so will reject theoretical models, since they don't have things like required REMARK lines and experimentally reasonable temperature factors.

	Ben Webb, Modeller Caretaker
--
             http://www.salilab.org/modeller/
Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage