Cambridge Structural Database Crystal Structures of Small Organic and Organometallic Compounds Dr. Clifford Felder, Dept. of Structural Biology, Weizmann Institute, 8 December 2003 I am presenting here only the older, “free” program questv5, which runs only in Unix. If you make a mistake, you cannot go back and fix it, but must run a new search. A newer search program, ConQuest, is also available. It can also run under MS Windows, and fixes this problem, but you must buy a separate license for every computer. The CSDS provides some auxilliary programs for more specialized functions, see their Web site for details. Two Kinds of Searches: 1) Specific Compounds : ligands and drugs to dock into protein receptors, initial coordinates for X-ray refinement programs, literature checks. 2) Geometric Properties : structures of specific functional groups, bond lengths and angles, non-bonded interactions, statistical and correlation studies.

Searches for Specific Compounds and Structures: Initial Cartesian Coordinates of Ligands or Drugs: - to dock into protein receptor active sites - for X-ray structure refinement or molecular mechanics (Note: the X-ray conformation may not be correct in this context!) Obtain Cartesian Coordinates for X-ray structures Published in the Literature - Literature checks of published structures Geometric Property Analyses: Accurately Characterize a Functional Group or Groups: - Tabulate All or Selected Bond Lengths and Angles - Degree of Planarity - Rotation about Bonds: cis vs. trans, 3-fold, freely rotating Nonbonded Interactions: - Close contact distances - Charge-charge and hydrophobic interactions - Hydrogen bonding - Cation-pi, pi-pi and other kinds of interactions - Define centroids, vectors and planes Analize Distributions of, and Correlations between Properties.

To use the CSD, you must log into a Unix or Linux (sorry, not Mac OSX) computer that has the CSD disk mounted, either directly, or by remote login from another Unix/Linux computer, a Mac running OSX from an X-terminal running under X11, or Mac running Classic or PC running MS Windows and an X11 emulation (MacX from ‘macserve’, inquire from WICC for Windows). Be sure your userid on the server has the environment variable and path set as above. Note that the WICC Unix machines do NOT link the CSD disk. Before logging into to the Unix server, type: hostname, and then xhost + server-name. After logging into the server, type setenv DISPLAY client-name.weizmannn.ac.il:0.0 > questv5 probname > > term x > menu full click with mouse Construct search question: Search - textual searches * Text - compound, authors, journal, formula * Numerical - year, unit cell parameters, R-factor Build - search fragment * 2-D Constrain - structural (bonding) constraints * 3-D Constrain - geometric (spatial) constraints -> Define Structure Quest - define search question, files saved, etc. -> Start-search - run the search * 2D/3D - displays each hit’s structural diagrams * 1D - displays its bibliographic information Setting up the Unix/Linux server: Insert into file /etc/fstab: mount morena:/public/csd /chem2/csd OR mount genewise:/databases/csd /chem2/csd Insert into every user’s.cshrc file: setenv CSD /chem2/csd/cambridge ; set path = ($path $CSD/bin)

In the Build menu, to draw a structure, select draw and draw in the box. To cause a break between bonded atoms, select move and then draw again. Then select individual atoms or bonds to change their type; you can also select a range of bond (VBT) or element types (ELDEF). To do textual searches, select the Search menu, and then Text (alphanumeric) or Numeric. Select an item with the mouse, type in the value at the bottom, and verify. Return to the Build menu to modify your search fragment. In the 2-D Constrain menu, you can specify the minimum or total (eg. coordi- nation) number of bonds to a given atom, number of hydrogens (normally hydrogens are ignored in search questions) and whether the given atom is part of a cyclic structure. In the 3-D Constrain menu, you can ask to tabulate the values of specific interatomic distances (bonds), angles and torsions. You can also specify ranges of values. But you can also specify novel geometric properties by defining centroids, bisecting and normal vectors and planes with respect to atoms, and setup and tabulate geometric properties based on these constructs. While there is also a special nonbonded interactions feature, it is very slow. When you are done, select Define Structure from the Build Menu, type Y to acccept the 3-D constraints (if any), and then select Quest. Then select the AND or OR boolean operators and those specific search fragments and/or textual searches you want to include in your search question. You also select which output files you want to keep (select FDAT or Mol2 if you want atomic Cartesian coordinates). Finally, select Start-Search to begin the search.

The 2D/3D menu lets you see both the structural and 3-D diagram of each found structure. Use the mouse and the rotation-translation controls in the lower right to rotate or translate the latter. The 1D menu displays the bibliographic and crystal data, much of which also appears in the console terminal window. Only once you are satisfied that you are getting the hits you want, you can select Diag or Nodiag to select everything hit, either showing diagrams (so you can monitor the search) or not (which is faster). Once the search is over, be sure to enter ‘Y’ to be sure to save your search results.

When done, you get a series of files: probname.jnl - search question and bibliographical references probname.dat - FDAT crystal data (readable by some programs, can also output MOL2 coords.) probname.gcd - list of REFCODES probname.sum - summary of search fragment results.fgd,.tab - files for VISTA Built-in help in almost every screen. Warning: subsequent jobs will over- write your files, use unique names. Auxiliary programs include: vista - analysis of geometric statistical results On-line help: /~cffelder/weiz/3dchem.html /~cffelder/weiz/csd-info.html Cambridge Database Home Page: Auxilliary program ‘babel’ can convert fdat to pdb format: babel -ifdat probname.dat -opdb probname.pdb phone 3077