Descriptor Generation

Input CSV:

SMILES	code_name
C	mol_1
CC	mol_2
CCC	mol_3
CCCC	mol_4

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    input="FILENAME.csv"
)

Outputs:

CSV files with descriptors (standard: AQME-ROBERT_interpret_FILENAME.csv)
No common substructure → only molecular descriptors generated

Input CSV:

SMILES	code_name
P	mol_1
PC	mol_2
P(C)C	mol_3
P(C)(C)C	mol_4

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    input="FILENAME.csv"
)

Outputs:

CSV files with descriptors
Common substructure detected (P atom)
Molecular and atomic descriptors generated

Input CSV:

SMILES	code_name
C(O)N	mol_1
CC(O)N	mol_2
CCC(O)N	mol_3
CCCC(O)N	mol_4

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    input="FILENAME.csv",
    qdescp_atoms=["O"]
)

Outputs:

Only O atoms used for atomic descriptors
Molecular descriptors always included

Note

Other SMARTS patterns can be used (e.g., C=O, C#N). The pattern must appear once in all molecules.

Input CSV:

SMILES	code_name
[P:1]([H])([H])([H])	mol_1
[P:1]([H])([H])C	mol_2
[P:1]([H])(C)C	mol_3
[P:1](C)(C)C	mol_4

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    input="FILENAME.csv",
    qdescp_atoms=["1"]
)

Outputs:

Descriptors only for atoms labeled with index 1
Molecular descriptors always included

Input CSV:

code_name	SMILES_phosph	SMILES_alk
mol_1	P	C
mol_2	PC	CC
mol_3	P(C)C	CCC
mol_4	P(C)(C)C	CCCC

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    input="FILENAME.csv"
)

Outputs:

Descriptors generated independently for each component and combined into a single descriptor array

Input CSV:

code_name	SMILES	charge	mult
mol_1	[H][N+]1([H])[Cu][N+]([H])([H])CC1	2	2
mol_2	[H][N+]1([H])[Cu]N([H])CC1	1	2
mol_3	[H]N1[Cu]N([H])CC1	0	2
mol_4	C[N+]1(C)[Cu][N+](C)(C)C=C1	2	2

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    input="FILENAME.csv"
)

Outputs:

Atomic descriptors for Cu automatically generated
Molecular descriptors always included
Charges and multiplicity required

Input CSV:

code_name	SMILES	mult	complex_type	geom
mol_1	Cl[Pd]([PH3+])(Cl)[NH3+]	1	squareplanar	"['[Cl][Pd][Cl]',180]"
mol_2	Cl[Pd](Cl)([P+](C)(C)C)[NH3+]	1	squareplanar	"['[Cl][Pd][Cl]',180]"
mol_3	Cl[Pd]([PH3+])(Cl)[N+](C)(C)C	1	squareplanar	"['[Cl][Pd][Cl]',180]"
mol_4	Cl[Pd](Cl)([P+](C)(C)C)[N+](C)(C)C	1	squareplanar	"['[Cl][Pd][Cl]',180]"

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    input="FILENAME.csv"
)

Outputs:

Atomic descriptors for Pd, N and P (detected automatically)
Molecular descriptors always included
Charges and multiplicity required
Squareplanar geometry enforced via the complex_type column
Two chloride ligands in trans (Cl-Pd-Cl angle at 180 degrees) enforced via the geom column

Input CSV:

code_name	SMILES	mult
mol_1	O=CC1=CN(C)[Cu]N1C	2
mol_2	O=CC1=C(C)N(C)[Cu]N1C	2
mol_3	O=CC1=C(N([Cu]N1C(C)(C)C)C)C	2
mol_4	O=C(C)C1=C(C)N([Cu]N1C)C	2

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    input="FILENAME.csv",
    qdescp_atoms=["C=O"]
)

Outputs:

Atomic descriptors only for C and O atoms from carbonyl groups
Molecular descriptors always included

Input CSV:

code_name	SMILES	charge	mult
mol_1	[H][N+]1([Cu][N:1](C=C1)[H])[H]	1	2
mol_2	[H][N+]1([H])[Cu][N:1]([H])CC1	1	2
mol_3	[H][N+]1([Cu][N:1](C(C)=C1)[H])[H]	1	2
mol_4	[H][N+]1([Cu][N:1](C(C)=C1C)[H])[H]	1	2

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    input="FILENAME.csv",
    qdescp_atoms=["1"]
)

Outputs:

Atomic descriptors only for indexed atoms
Molecular descriptors always included

Python (Jupyter Notebook):

from aqme.qdescp import qdescp

qdescp(
    files="*.sdf"
)

Outputs:

CSV descriptor files generated from 3D structures

Note

Useful when SMILES are not suitable (e.g., axial chirality or noncovalent complexes).