Tag SMARTS query hits with their labels/names

For a given input structure file I like to tag it with another list of predefined SMARTS queries. The SMARTS queries are stored in a file, and it has following format:

SMARTS1, name1

SMARTS2,name2

If a given smarts matches a target mol tag the mol with the smarts name, if it matches multiple smarts concatenate the smarts names to create the tag; if a mol does not match any smarts then send the mol out without tagging.

I am looking for a simple command line solution for this, and have checked the jcsearch and evaluate, but did not quite know how to use these commands to accomplish the task. A simple JAVA code would work as well.

Thanks in advance

Hi,

Do you happen to be using Pipeline Pilot ?

If yes, there might be a solution W/O coding by using our components for PP.

(probably KNIME would also work in a similar way as well)

If not using either of these then some custom Java code may be needed.

 

Best regards,

Szilard

no, i do not use pp, why do i need that. is it possible to do this using evaluate?

dlee wrote:

no, i do not use pp, why do i need that. is it possible to do this using evaluate?

I think if you are not familiar with the product already it probably does not justify the effort to set up this infrastructure just to accomplish this one task - writing a simple Java program can be quicker.

We have prepared a simple Java program that does the job (attached)

Best regards,

Szilard

Thanks, that is all need, simple and efficient

Sziliard:

Sorry to take up on a old post, but I noticed you mentioned using pipeline pilot in order to search a set structures using a list of SMARTS, could you provide any insight on this?  

Much Thanks

Dennis

In PLP there is a component called 'Substructure Map from Tag'. What it does is to read in a stream of molecules which can be a mix of the molecules to be searched and the query molecules which can be smarts. You need to convert the smarts into molecules using convert component, and lable them as 'IsQuery' so the map component knows which are the target or query mols. The map component has whole bunch of the parameters to control its behavior  such as if you wish to tag all the matchings you can specify keep going do not stop after the first matching, and the name of the matching substructures (smarts) are append into an array variable called QueriesMapped.

Thank you very much, that is very helpful

 

Dennis

Szilard wrote:

dlee wrote: no, i do not use pp, why do i need that. is it possible to do this using evaluate? I think if you are not familiar with the product already it probably does not justify the effort to set up this infrastructure just to accomplish this one task - writing a simple Java program can be quicker. We have prepared a simple Java program that does the job (attached) Best regards, Szilard

 Hi Szilar,

I downloaded your code to my lap top and try to compile it, but failed. I am pretty sure it is related to classpath etc. I am using following line to compile your code:

./javac -classpath "/cygdrive/c/Program Files/ChemAxon/JChem/lib/;../lib/;../jre/lib" ./TagByQuery.java

I am using cygwin, my current working dir is :

/cygdrive/c/Program Files/Java/jdk1.6.0_21/bin

The main error is:

.\TagByQuery.java:1: package chemaxon.formats does not exist
import chemaxon.formats.MolImporter;
                       ^

Thanks in advance

Dong

 

 

Hi Dong,

Yes, the problem is in the classpath.

1. Your classpath exhibits a more general cygwin usage problem: programs which don't come with cygwin won't understand
   
   
   
   1. either the unix style paths in general (slashes and colons -- I realize now that you're mixing slashes [unix directory separators] with semi-colons [windows path separators])
   
   
   2. or cygwin's represtation of absolute paths in particular. 
   
   
   "Native" Windows programs expect paths in native Windows format: C:\Program Files\ChemAxon . (Don't forget to use single quotes then...). Alternatively you can use the cygpath program to convert unix style paths to Windows-style. (If you so little experienced with cygwin, you'd probably be much better off using plain old Command Prompt...)
   


2. There is also a Java-specific problem with your classpath. If a Java classpath element points to a directory, that directory must be the parent of a class file hierarchy. In contrast to class files, JAR files must be distinct elements in Java classpaths -- JAR files will not be considered, if only their parent directory appears in the classpath.

Peter