GenerFP

Version 5.9.4

Notice: This program is obsolete. Please use GenerateMD instead.

GenerFP is a Java program for generating binary fingerprints.

Contents

• What are chemical hashed fingerprints?
• Requirement
• How to use GenerFP?
• How does it work?
• Examples

 

What are chemical hashed fingerprints?

A chemical hashed fingerprint is a bit string generated from the structure of a molecule. The number of different bits in the fingerprint of similar molecules is lower than in the case of dissimilar ones. GenerFP uses a hashing algorithm to generate fingerprints from molecules that can be used for

• quantifying the similarity of molecules
• predicting the diversity of a set of compounds
• searching substructures in molecules.

 

Requirement

The GenerFP program is written in Java. To run GenerFP Java Virtual Machine 1.1.6 or later is needed.

 

How to use GenerFP?

Usage

    generfp [<options>] <inputfile >outputfile

Prepare the usage of the generfp script or batch file as described in Preparing the Usage of JChem Batch Files and Shell Scripts.

Or call the GenerFP class directly:

Win32 / Java 2 (assuming that JChem is installed in c:\jchem):

    java -cp c:\jchem\lib\jchem.jar chemaxon.sss.screen.GenerFP [<options>] <inputfile >outputfile

Unix / Java 2 (assuming that JChem is installed in /usr/local/jchem):

    java -cp /usr/local/jchem/lib/jchem.jar chemaxon.sss.screen.GenerFP [<options>] <inputfile >outputfile

Options

  -h               display this help and exit
  -fl <length>     fingerprintlength length in bytes (default: 64)
  -pl <length>     maximum number of bonds in patterns (default: 5)
  -bc <bitcount>   number of bits to switch on for each pattern
                    in the structure (default:2)
  -f<format>       format of the output
     -fb              binary
     -f1              ones and zeros  (001011011...) (default)
     -fd              bytes in decimal format
     -fh              bytes in hexadecimal format
     -fi              integers in decimal format

  -stat            generate statistics

  -s <separator>   separator in the output in case of text output
      Separators:     
	    n no separator
            c comma (default)
            t tab
            s space

Input file formats

• MDL molfile
• Compressed molfile
• SDfile
• Compressed SDfile
• SMILES

 

How does it work?

For each atom of the molecule the following algorithm is performed:

1. Finding all path from the given atom, which can be accessed through 0, 1, 2,...<length> bonds in the structure.

The -pl <length>  option sets the maximum length of path.
2. Generation bit sets for paths.

For each of the previously generated paths, a bit array is generated with fixed length and fixed number of bits turned on.
The -fl <length> option sets the fixed length of the bit array.
The -bc <bit count> option sets the number of bits to turn on for each path.
3. Add the bit sets to the fingerprint using logical OR operation.

Generating fingerprint for CH₃-CH₂-OH

• Find the paths up to 3 bonds:

For the 1^st C atom:


0 - bond	C
1 - bond	C-H,C-C
2 - bond	C-C-H,C-C-O
3 - bond	C-C-O-H

And for all the other atoms...
 

• Generate a bit array with length 10 and 2 bits turned on:



path	bit array
C	1010000000
C-H	0001010000
C-C-H	0001000010
C-C-O	0100010000
C-C-O-H	0000001001

And for all the other paths...
 

• Add these bit arrays to the fingerprint:



0000000000	initial fingerprint
1010000000	1-st path
0001010000	2-nd path
0001000010	3-rd path
0100010000	4-th path
0000001001	5-th path
1111011011	partial result


Continue for all the other bit arrays....

The partial result is: 1111011011 .

 

Examples

The following examples work under Unix.

• generfp.sh -h


Shows the help screen.



• generfp.sh <input.mol >out.txt


This generates the fingerprint for the input molfile with the default settings:


Fingerprint length: 64 byte
Number of bits turned on for each pattern: 2
Maximum number of bonds to use for generate patterns: 5
Output format: ones and zeros
The result out.txt is a text file.


out.txt:
0,1,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,... (64*8 number)


• generfp.sh -fl 64 -bc 1 <input.sdf >out.txt

The above line generates 64 byte long fingerprints using patterns containing maximum 5 bonds, for each molecule in the input file. 1 bit will be generated for each pattern in the hashing procedure.

out.txt:
1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,... (64*8 number)



• generfp.sh -fh -s s <input.mol >out.txt

Generates the fingerprint for the molfile with the default settings. The result is in hexadecimal format separated by spaces in the out.txt file.

out.txt:
8 0 6 0 8 a 0 4 6 8 0 0 0 0 1 0 0 0 ... (64*8 char)



• generfp.sh -stat <molecules.sdf

Generates statistic for the input structure file indicating the

• Average number of 1 bits (the average number of 1 bits in the fingerprint)
• Maximum number of 1 bit and the id. number of the molecule at the maximum
• Minimum number of 1 bits and the id. number of the molecule at the minimum
• Density function

output:



Number of bits(1) set on:
Average: 7.12%
Maximum: 51.89% (at molecule 1206)
Minimum: 0.20% (at molecule 1201)
Density function:
     0%-10%   79.53%
    10%-20%   16.80%
    20%-30%    2.58%
    30%-40%    0.47%
    40%-50%    0.57%
    50%-60%    0.05%
    60%-70%    0.00%
    70%-80%    0.00%
    80%-90%    0.00%
    90%-100%   0.00%