From: Chris M. <c.m...@ds...> - 2007-08-31 19:12:41
|
Craig A. James wrote: > Chris Morley wrote: >>>> Another question is whether the mean of each bit is fixed or changed >>>> according to the input molecules. >>> I think you're asking, "the meaning", that is, what each bit >>> represents. The answer is no, the meaning of each bit does not ever >>> change, no matter what molecule is input. However, there is no way >>> to know what each bit means since a hash is used. >>> >> The default fingerprint in OpenBabel, "FP2", is of this Daylight type >> as above. >> >> "FP4" is of a different type. Each bit represents a chemical feature >> defined by a set of 307 SMARTS patterns... > > Ok, I know this is a quibble, but... > Strictly speaking, shouldn't these be called "structural keys" rather > than "fingerprints"? Even though FP4 uses OB's fingerprint object, it > really should be called something like "SK4". (I suppose the > fingerprint object should have a more generic name like "BitVector". > Not that we should change the C++ class names.) Many people are > introduced to cheminformatics through OpenBabel, and correct terminology > can be very helpful. > FP4 is just a name that the user has to remember. I think that FP4 is likely to be easier for the casual user than introducing an additional set of letters, however correct it is. But I agree with your general point. Names in programs get chosen without too much thought - at the time the program logic is much more interesting. They are then set in stone and often prove to be opaque to other users. Short of having a nomenclature committee to review programs shortly after they are written, and it is a very special kind of person who would want to take part in that, it is difficult to see what to do. More disciplined programmers would of course be a help... Chris |