For each record from the benzodiazepine file, print the total number of heavy atoms in each record (that is, exclude hydrogens). You must get the data from the connection table and not from the tag data. (In particular, don't use PUBCHEM_HEAVY_ATOM_COUNT .)

The output is one output line per record, containing the count as an integer. If at all possible, show how to read directly from the gzip'ed input SD file. The first 5 heavy atom counts are 21, 24, 22, 22 and 22. A copy of the expected output is available for reference.

Some of the hydrogens in the data set have atomic weights of 2 or 3, which may cause difficulties for toolkits which convert 1H hydrogens to implicit form while leaving 2H and 3H to explicit form. Please make sure your output agrees with the reference output.

Suggestion: Please add a couple of '$$$$' SD data lines to the file, to trip those naive implementations which think that a '$$$$' is automatically a record separator!

CDK/Java[edit | edit source]

Save the code to "" then do

   java -cp ./:$HOME/cdk- countHeavyAtoms > countHeavyAtoms.out
import org.openscience.cdk.interfaces.IChemObjectBuilder;
import org.openscience.cdk.Molecule;
import org.openscience.cdk.DefaultChemObjectBuilder;
import org.openscience.cdk.interfaces.IAtom;

public class countHeavyAtoms {	
    public static void main(String[] args) {	
        FileReader sdfile = null;
        try {
            /* CDK does not automatically understand gzipped files */
            sdfile = new FileReader(new File("benzodiazepine.sdf"));
        } catch (FileNotFoundException e) {
            System.err.println("benzodiazepine.sdf not found");
        IteratingMDLReader mdliter = new IteratingMDLReader(sdfile,
        while (mdliter.hasNext()) {
            Molecule mol = (Molecule);
            int numHeavies = 0;
            for (IAtom atom : mol.atoms()) {
                if (atom.getAtomicNumber() > 1) {

CDK/Groovy[edit | edit source]

Save the code to "countHeavyAtoms.groovy" then do:

 groovy countHeavyAtoms.groovy > countHeavyAtoms.out

import org.openscience.cdk.interfaces.*;
import org.openscience.cdk.silent.*;

iterator = new IteratingMDLReader(
  new GZIPInputStream(
    new File("benzodiazepine.sdf.gz").newInputStream()
while (iterator.hasNext()) {
  mol =
  heavyAtomCount = 0
  for (atom in mol.atoms()) {
    if (1 == atom.atomicNumber || "H" == atom.symbol) {
      // do not count hydrogens
    } else {
  println heavyAtomCount

CDK/Cinfony/Jython[edit | edit source]

from cinfony import cdk

manip =
for mol in cdk.readfile("sdf", "benzodiazepine.sdf"):
    print len(manip.getHeavyAtoms(mol.Molecule))

Chemkit/C++[edit | edit source]

#include <iostream>

#include <chemkit/atom.h>
#include <chemkit/foreach.h>
#include <chemkit/molecule.h>
#include <chemkit/moleculefile.h>

using namespace chemkit;

int main(void)
    MoleculeFile file("benzodiazepine.sdf.gz");
    bool ok =;
        std::cerr << "Error: " << file.errorString() << std::endl;
        return -1;

    foreach(const Molecule *molecule, file.molecules()){
        int heavyAtomCount = 0;

        foreach(const Atom *atom, molecule->atoms()){

        std::cout << heavyAtomCount << std::endl;

    return 0;

Indigo/C++[edit | edit source]

#include "base_cpp/scanner.h"
#include "molecule/molecule.h"
#include "molecule/sdf_loader.h"
#include "molecule/molfile_loader.h"

int main (void)
   const char *filename = "benzodiazepine.sdf"; // you can use ".sdf.gz" as well

      FileScanner scanner(filename);
      SdfLoader loader(scanner);

      while (!loader.isEOF())
         Molecule mol;
         BufferScanner molscan(;
         MolfileLoader molload(molscan);
         molload.loadMolecule(mol, false);
         int heavy_count = 0;
         for (int i = mol.vertexBegin(); i != mol.vertexEnd(); i = mol.vertexNext(i))
            if (mol.getAtom(i).label > ELEM_H)

         cout << heavy_count << endl;
   catch (Exception &e)
      cerr << "error: " << e.message() << endl;
      return -1;

   return 0;


  1. Unpack 'graph' and 'molecule' projects into some folder
  2. Create 'utils' folder nearby
  3. Paste the above code into utils/count_heavy.cpp file
  4. Compile the file using the following commands:
    $ cd graph; make CONF=Release32; cd ..
    $ cd molecule; make CONF=Release32; cd ..
    $ cd utils
    $ gcc count_heavy.cpp -o count_heavy -O3 -m32 -I.. -I../common ../molecule/dist/Release32/GNU-Linux-x86/libmolecule.a ../graph/dist/Release32/GNU-Linux-x86/libgraph.a -lpthread -lstdc++
  5. Run the program like that:
    $ ./count_heavy

Indigo/Python[edit | edit source]

from indigo import Indigo
indigo = Indigo()

for mol in indigo.iterateSDFile("benzodiazepine.sdf.gz"):
    print mol.countHeavyAtoms()

OpenBabel/Python[edit | edit source]

import openbabel as ob

obconversion = ob.OBConversion()

obmol = ob.OBMol()
notatend = obconversion.ReadFile(obmol, "benzodiazepine.sdf.gz")
while notatend:
    print obmol.NumHvyAtoms()
    # reuse the molecule for 0.3 seconds of savings vs. making an new one
    notatend = obconversion.Read(obmol)

OpenBabel/Pybel[edit | edit source]

This builds on and simplifies OpenBabel

import pybel

for mol in pybel.readfile("sdf", "benzodiazepine.sdf.gz"):
    print mol.OBMol.NumHvyAtoms()

OpenBabel/Python[edit | edit source]

import openbabel as ob

obconversion = ob.OBConversion()

obmol = ob.OBMol()
notatend = obconversion.ReadFile(obmol, "benzodiazepine.sdf.gz")
while notatend:
    print obmol.NumHvyAtoms()
    # reuse the molecule for 0.3 seconds of savings vs. making an new one
    notatend = obconversion.Read(obmol)

OpenBabel/Rubabel[edit | edit source]

require 'rubabel'
Rubabel.foreach("benzodiazepine.sdf.gz") {|mol| puts mol.ob.num_hvy_atoms }

OpenEye/Python[edit | edit source]

from openeye.oechem import *

ifs = oemolistream()"benzodiazepine.sdf.gz")
for mol in ifs.GetOEGraphMols():
    print OECount(mol, OEIsHeavy())
    # Can't use
    #    print mol.NumAtoms()
    # because the input structure contains explicit hydrogens.

RDKit/Python[edit | edit source]

I could also have iterated over the atoms and only counted the non-hydrogen ones, but I doubt that would be any faster or elegant.

from rdkit import Chem

# RDKit does not seem to support gzip'ed SD files
suppl = Chem.SDMolSupplier("benzodiazepine.sdf")
heavy_patt = Chem.MolFromSmarts("[!#1]")

for mol in suppl:
    print len(mol.GetSubstructMatches(heavy_patt))

The method "mol.GetNumAtoms()" is supposed to report only the non-hydrogen counts, but actually reports the number of explicit atoms in the molecular graph.

RDKit/Cinfony/Python[edit | edit source]

from cinfony import rdk

for mol in rdk.readfile("sdf", "benzodiazepine.sdf"):
    print mol.calcdesc(['HeavyAtomCount'])['HeavyAtomCount']

Unix command-line[edit | edit source]

It is possible to do this on the Unix command-line using a single (very long) line. To make it easier to understand I've written it using an awk script named 'heavy_atom_counts.awk'. Use it as:

gzcat benzodiazepine.sdf.gz | awk -f heavy_atom_counts.awk
# This file is named "heavy_atom_counts.awk"
# Check track of the line number in the record
{lineno += 1}

# Get the atom counts from the record
lineno == 4 { num_atoms = int(substr($0, 1, 3)); num_hydrogens = 0}

# Count the number of hydrogens
4 < lineno && lineno <= 4+num_atoms && substr($0, 32, 2) == "H " {num_hydrogens++}

# And the end of the record, print the number of heavy atoms and reset
/^\$\$\$\$/ {print num_atoms - num_hydrogens; lineno = 0}

Cactvs/Tcl[edit | edit source]

molfile loop "benzodiazepine.sdf.gz" eh {
   puts [ens get $eh E_HEAVY_ATOM_COUNT]

The toolkit can read compressed files directly. The loop construct automatically discards the structure read for each record, so this works on arbitrarily big files. Also, the format of the input file can be anything which can be identified automatically, so this is not limited to SD files.

Cactvs/Python[edit | edit source]

for eh in Molfile('benzodiazepine.sdf.gz'):

The Python language version has the same capabilities as the Tcl variant.

Community content is available under CC-BY-SA unless otherwise noted.