[Biophp-dev] Zend and Parser-to-SQL scripts

Serge Gregorio biophp-dev@bioinformatics.org
Sun, 08 Jun 2003 14:53:39 +0800


>> Serge: Btw, is "Zend Accelerator" = "Zend Compiler"?

> Nico: One last try:

Hehehe, sorry for trying your patience.  =)

>Nico: We are talking about two different things.  

>One is the ability to extend the functionality of 
>php.  <SNIP> You could simply take the clustalw >source code, <SNIP> compile php together with your >code, and voila, you >now have your function ...
>which behaves exactly like any... php function. 

>The other subject is acceleration of php.  The Zend >accelerator (or whatever they call it) stores a >compiled version of your php script in memory or on >disk.  This is worthwhile mainly for heavy duty >servers (quite a few cpu cycles will be avoided).  

>I have the feeling you are mainly interested in >option One.  Go ahead, check it out, you probably >have all the tools you need already...

Ok, thanks for the clarification. Hmm, both would
be useful to our project, but I'm more interested
in the second one. I've heard some complaints that say, BioPython is slow at parsing large files
(e.g. GenBank with thousands of records/entries). 

Though I haven't really tested our parsers against large datasets, I wouldn't be surprised if they 
turn out to be too slow for some users.  Thus, the
idea of compiling (some) BioPHP code came to mind.

The first option you were mentioning is something Cayte might be interested in, am I right Cayte? 

>> Serge: Yes.  Initially, the code would write to
>> some PRE-DESIGNED set of tables.  Later, we can let >> the user design his own database for, say, 
>> holding GenBank DNAs or PDB/mmCIF protein records.  

>> Hmm... is there any open-source tool out there that >> does this?  Oh well, if there isn't, we can always >> write one.

> Nico: Ehmm., shouldn't we write the php scripts that > write a seqobject to an SQL database (with the 
> structure I am getting into the adodb script right 
> now)?

Well, yes, that could be starting point.  But in my last paragraph, I was already thinking beyond that (having a fixed database design).  I was thinking of a tool/script/webpage where users can decide what their tables look like, and our Seq object-to-SQL code generator, would easily adapt to their choices, i.e.
generate INSERT statements that conform to their chosen table structures/design.



Need a new email address that people can remember
Check out the new EudoraMail at