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

Re: [modeller_usage] Regarding Scoring Parameters



I am not sure I understand your question, and I will assume that by "normal set of these parameters" you mean "unifying set across different programs". If that's your question, the answer is that there is no common set of lambda and kappa parameters (extreme value distribution factors) between different programs, or even within the same program. The values of these parameters are dependent on the scoring scheme used, so when you change the substitution matrix in PSI-BLAST, they will change. If you calibrate HHPRED models in "local" and "global" modes, you'll get different kappa and lambda values, and the same is true if you change other program parameters that affect the scoring and alignments. If you take a sequence (or profile) with certain composition and length and calculate lambda and kappa, they will be different from the next sequence depending on its length and sequence composition. In an extreme case, you can even take the same HHPRED profile and calibrate it against two different databases, for example the one provided with the program, and another database with similar number of profiles, and you'll most likely get different extreme value distribution parameters. Before anyone freaks out because of this uncertainty, it bears mentioning that most of these changes would affect only borderline matches.

This is to say that you can't determine these parameters in one program and then use them with others. You can't even determine these parameters for local scoring and then use them for global scoring with the same program. In short, extreme value distribution parameters are both sequence-, program- and scoring parameter-specific.

Hope this helps,

Mensur


At 12:16 AM 2/9/2012, you wrote:
Dear Modellers,
For E-value and alignment scoring systems, we use the parameters lambda and K. Specifying these scores is must for a scoring system. So what is/are the normal set of these parameters used in PSI-BLAST/CS-BLAST/HHPRED or other profile inclusion based template searching methods? Tell me the commonly used values for these two parameters lambda and K for the calculation of alignment scores.

Thanks
Ashish

Ashish Runthala,
Lecturer, Structural Biology Cell,
Biological Sciences Group,
BITS, Pilani
Rajasthan, INDIA
_______________________________________________
modeller_usage mailing list

https://salilab.org/mailman/listinfo/modeller_usage

==========================================================================
| Mensur Dlakic, PhD                | Tel: (406) 994-6576                |
| Department of Microbiology        | Fax: (406) 994-4926                |
| Montana State University          |                                    |
| 109 Lewis Hall, P.O. Box 173520   | http://myprofile.cos.com/mensur    |
| Bozeman, MT 59717-3520            | E-mail:         |
==========================================================================