Quality control (QUALTY)

Introduction.

The first obvious question is of course "What is quality control?".

To answer this we need a little bit of background information. WHAT IF has the possibility to analyse atomic contacts of any desired kind in the whole PDB database. For example, the SCNCON and SCNGRN option in the SCAN3D menu allow you to get all tyrosine - aspartic acid contacts in the whole PDB up at the screen just a few seconds CPU. WHAT IF does the following: It puts one tyrosine at the screen, this is called the central residue. It will then loop over all tyrosines in the PDB. Each of those wil be superimposed on the central one. The matrix needed to superimpose a tyrosine on the central one will be applied to the whole molecule. Subsequently, all aspartic acids in the whole molecule that make a contact with the tyrosine that is superimposed on the central one will be drawn.

It is obvious that if there are many contacts, such a plot becomes rather crowdy and unclear. It would be much nicer to have some kind of a probability function in space that can be contoured along equi-probability lines, similar as is done with electron density by crystallographers. Such a function should then have a higher value at positions where more atoms are falling on top of each other. Quality boxes provide such a function. WHAT IF has probabilty boxes available for all atom types around all residues.

In the above a tyrosine was taken because it is rather obvious how to superimpose two tyrosine side chains on top of each other. For example with lysine this is much less obvious, and it is clear that several runs, each time superpositioning on another set of (minimally) three atoms is needed. These sets that can be used for superpositioning are called maximally rigid sub-structures.


There are two kinds of options in the quality menu: Useful ones, and useless ones, the useless options are activated by the MORE command. The latter category are the options required to generate and inspect the quality control boxes.

The text of an article written on this subject is added to this writeup as an appendix.


This menu will be rewritten in early 1995

Usage of quality control boxes

Testing a range of residues (RNGQUA)

The command RNGQUA will cause WHAT IF to prompt you for a range of residues. The deviation from the average quality value for all residues in this range will be shown. Depending on the parameter setting you will or will not see which atoms contributed in which way to the values per residue. Since this option needs the surface contact area to be known, you will be brought in the accessibility module if this has not been done prior to calculating quality values.

Quality control is a time consuming option. If you do not have a log-file open, you will be asked if you really want to continue without log-file.

In an individual residue has a quality control value of -5.0 or worse, you should take a look at it. It means:


It is involved in symmetry contacts, or
It is binding a co-factor, ligand or ion, or
It is an active site residue, or
It is wrong.

Average protein values can be read as follows:

Greater than -0.5 perfect
-0.5 Average good protein
-1.0 till -0.5 still good or very good model
-1.5 probably still OK, but with many small errors, normal model
-2.0 A ab initio designed protein or very poor real protein
-3.0 Guaranteed wrong protein or bad model

Predicting mutants (MUTQUA)

The command MUTQUA will cause WHAT IF to prompt you for a residue range. Due to the special nature of this process, the terminal residues of a molecule can not be included in this range. (You can of course fudge this with the PASTE command in the SOUP menu, but then I give no guarantees about what is going to happen).

WHAT IF will then try the 20 possible amino acids at every given position. It will for every mutant find a `likely` side chain conformation, and calculate the quality value for the mutant in this position. Because of the fact that only one likely position will be tested, it can happen that the original residue after mutating via several other residues back into itself gives a different value. In cases where this value is higher than the original value, WHAT IF thinks that it can improve nature. This is of course a good indicator for the unreliability of this process.

Quality control is a time consuming option. If you do not have a log-file open, you will be asked if you really want to continue without log-file. You will also be asked if the packing quality for the residues should be stored in a table (See menu on TABLES).

Quality control parameters (PARAMS)

The command PARAMS brings you in the menu from which you can change the QUALTY menu related parameters.

the following parameters can be set:

LEVOUT   Level of output in quality control (0-2).
         The higer the level, the more output
WAYQUA   Way in which the quality control values are calculated.
       0 Take only the nearest point in the qulaity control box
       1 (Default) convolute the atom with the quality box
       2 Average out the directionallity of the contac boxes
SKPNEI   Flag for skipping inter residue covalent contacts
       0 Skip nothing
       1 Skip mainchain atoms of covalent neighbour
WEIGTH   Weigth used when adding up neighbours contributions
       0 (Default) Use full neighbour and set weight to shortes
         contact distance between fragment and neighbour for the
         whole neighbour residue.
       1 Only use contributions of contacting atoms weighed by 
         the individual shortes contact.

Showing parameters (SHOPAR)

The command SHOPAR will show you the present value of the parameters. See PARAMS above for the list of parameters available.