From: Mgr. Lubos Vrbka (lubos.vrbka_at_uochb.cas.cz)
Date: Mon Apr 18 2005 - 09:46:40 CDT

hi,

> ...
> Having on-the-fly decompression for individual structures stored in
> text files is probably a nice feature to have, so if John's to-do list
> is too short, maybe he can add this :-)
in my opinion vmd handles every file that is loaded as a trajectory -
some with single frame only... so such a feature will influence also the
trajectories

> There are some text structure file formats that can contain several
> structures and these can be used as a sort of trajectories; also,
> CHARMM for example has the option of converting a binary trajectory
> into a text one, with precision loss - these text files can be
> compressed much better by gzip, but the text files are bigger than the
> binary equivalents, so per-ensemble the gain is not worth it.
>
regarding the space saving: huge amount of space can be saved by using
e.g. text amber trajectory. in both binary and text format 1 coordinate
occupies 8 bytes (8.3 format in text). you are right with compression of
binary files - it is usually poor... but for text, it is different
story. of course, processing text files is much slower, but there are
also some advantages of them.

you are right that there can be some problems with binary files. i am
not sure, but isn't there some header in gzip'ed files? i think that
functions from zlib somehow detect the format and if the file is
"plain", then classical i/o routines are used.

other possibility - add "compressed" button to file selector window, so
the user can explicitly say - this is compressed...

regards,
lubos

-- 
.....................................................
Mgr. Lubos Vrbka
Center for Biomolecules and Complex Molecular Systems
Institute of Organic Chemistry and Biochemistry
Academy of Sciences of the Czech Republic
Prague, Czech Republic
.....................................................