Chemoinformatics recipes

Command line recipes for the working chemoinformatician

Unique InChI filtering of molecules (remove duplicates)

obabel INFILE -O OUTFILE --unique

Assign MMFF94 partial charges

obabel INFILE -O OUTFILE --partialcharge mmff94

Assign MMFF94 partial charges and hydrogens (protonate) at given pH (7.4) usinb openbabel

obabel in.smi -O out.mol2 --partialcharge mmff94 -p 7.4

Major tautomer at pH 7.4 usinb ChemAxon cxcalc

(-g: ignore errors; -H: pH; -f sdf: force output format to sdf, to preserve molecule names)

input file not being a .smiles might crash the tool!

cxcalc -g majortautomer -H 7.4 -f sdf input.smiles > output_taut74.sdf

Compute MACCS 166bits fingerprints and output them as strings

(will create a .csv file named after the input file)

mayachemtools/bin/MACCSKeysFingerprints.pl --size 166 [INFILE] --CompoundIDMode MolName

3D conformer generation using Corina classic

(one low energy conformer per molecule)

the optional [-d wh] add/writes out hydrogens (makes them explicit)

corina [-d wh] < INPUT.sdf > OUTPUT.sdf

lowest energy conformer generation using cxcalc from Chemaxon

cxcalc conformers in.smi -m 1 > out.sdf

lowest energy conformer generation using omega from OpenEye scientific

omega -in in.smi -out out.sdf -maxconfs 1

RMSD between two ligands (curr. will be superposed onto ref.)

fconv -rmsd current.mol2 --s=reference.mol2

compute Bemis-Murcko scaffolds of molecules

stripper --in molecules.smi --out scaffolds.txt

print a molecule in EPS format (for LateX manuscripts); obabel then inkscape

SMILES to EPS, MOL2 to EPS or SVG to EPS would work the same

obabel molecule.smi -O molecule.svg
inkscape molecule.svg -E molecule.eps --export-ignore-filters --export-ps-level=3

smi2eps in Bash (smi -> svg -> pdf -> cropped-pdf -> ps -> eps)

# librsvg2-bin provides rsvg-convert
# texlive-extra-utils provides pdfcrop
# ghostscript provides pdf2ps
# ps2eps provides ps2eps
function svg2eps () {
    tmp_pdf_out=`echo $1 | sed 's/\.svg$/\_tmp.pdf/g'`
    pdf_out=`echo $1 | sed 's/\.svg$/\.pdf/g'`
    ps_out=`echo $1 | sed 's/\.svg$/\.ps/g'`
    eps_out=`echo $1 | sed 's/\.svg$/\.eps/g'`
    svg=$1
    rsvg-convert -f pdf $svg -o $tmp_pdf_out
    pdfcrop $tmp_pdf_out $pdf_out
    pdf2ps $pdf_out $ps_out
    ps2eps < $ps_out > $eps_out
}

# openbabel provides obabel
function smi2eps () {
    smi=$1
    svg_out=`echo $1 | sed 's/\.smi$/\.svg/g'`
    obabel $smi -O $svg_out -xC -xd
    svg2eps $svg_out
}

Install open babel from sources

wget https://github.com/openbabel/openbabel/archive/openbabel-2-4-1.tar.gz
tar xzf openbabel-2-4-1.tar.gz
cd openbabel-openbabel-2-4-1/
mkdir build
cd build
cat <<EOF > build.sh
mkdir -p ~/usr
cmake -DPYTHON_BINDINGS=true -DCMAKE_INSTALL_PREFIX:PATH=$HOME/usr ../
EOF
chmod 755 build.sh
./build.sh
make -j4
make install

Install rdkit on Mac OS X:

brew tap rdkit/rdkit
brew install rdkit --with-python3 --with-inchi

If this does not work, try the conda way (but then usage will need to be in a conda environment):

wget -c https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-x86_64.sh
sh Miniconda3-latest-MacOSX-x86_64.sh -p ~/usr/miniconda3
~/usr/miniconda3/bin/conda install -q -y -c conda-forge rdkit

Now you should check that you can really use it from Python:

python3
import rdkit
from rdkit import Chem
m = Chem.MolFromSmiles('n1ccccc1')

Count molecules, works for various file formats

Store this in a 'molcount' script, somewhere on your PATH.

#!/bin/bash

for f in "$@"; do
    filename=`basename "$f"`
    extension="${filename##*.}"
    case "$extension" in
        mol2) egrep -c MOLECULE $f
              ;;
        plr) egrep -c '^END$' $f # position and contrib per atom to cLogP
             ;;
        pqr) egrep -c ^COMPND $f
             ;;
        sdf) grep -c '$$$$' $f
             ;;
        mol) grep -c '$$$$' $f
             ;;
        phar) grep -c '$$$$' $f # Pharao DB
             ;;
        smi) cat $f | wc -l
             ;;
        *) echo "molcount: unsupported file format: ."$f
           ;;
    esac
done

Get molecules by name, for various file formats

Works even with a "database" file with millions of molecules.

lbvs_consent_mol_get from https://github.com/UnixJunkie/consent

lbvs_consent_mol_get -i molecules.{sdf|mol2|smi} {-names "mol1,mol2,..."|-f names_file}

Sayle hashing of a molecule

Some kind of canonicalization of molecular representations, consisting in the pair:

Sayle_hash(m) = (Canonical_smile_forcing_only_single_bonds_and_noH(m), number_of_Hydrogens_on_non_carbons(m) - sum_of_formal_charges(m))

m being the molecule to hash.

install deepchem on a Mac or Linux

No GPU support, but at least its an automatic and simple install procedure. Deepchem's version is fixed to a version that works for what I currently do.

pip3 install joblib pandas sklearn tensorflow pillow simdna deepchem==2.1.1.dev353

standardize molecules in parallel with pardi and standardiser

pip3 install chemo-standardizer

opam install pardi

#!/bin/bash

if [ $# -lt 1 ]; then
    echo "usage: "$0" input.smi output_std.smi"
    exit 1
fi

INPUT=$1
OUTPUT=$2

pardi -i $INPUT -o $OUTPUT -c 400 -d l -ie '.smi' -oe '.smi' \
      -w 'standardiser -i %IN -o %OUT 2>/dev/null'

Links / Bibliography

[1] http://www.mayachemtools.org/

[2] http://openbabel.org/wiki/Main_Page

UnixJunkie / chemoinfo_recipes