#
#              SMARTS Patterns for Functional Group Classification 
#
#              written by Christian Laggner 
#              Copyright 2005 Inte:Ligand Software-Entwicklungs und Consulting GmbH
#
#              Released under the Lesser General Public License (LGPL license)
#              see http://www.gnu.org/copyleft/lesser.html
#              Modified from Version 221105
#####################################################################################################

# General Stuff:
# These patters were written in an attempt to represent the classification of organic compounds 
# from the viewpoint of an organic chemist.
# They are often very restrictive. This may be generally a good thing, but it also takes some time
# for filtering/indexing large compound sets. 
# For filtering undesired groups (in druglike compounds) one will want to have more general patterns 
# (e.g. you don't want *any* halide of *any* acid, *neither* aldehyde *nor* formyl esters and amides, ...). 
#

# Part I: Carbon 
# ==============


# I.1: Carbon-Carbon Bonds
# ------------------------

# I.1.1 Alkanes:

[CX4H3][#6] Primary_carbon

[CX4H2]([#6])[#6] Secondary_carbon

[CX4H1]([#6])([#6])[#6] Tertiary_carbon

[CX4]([#6])([#6])([#6])[#6] Quaternary_carbon


# I.1.2 C-C double and Triple Bonds

[CX3;$([H2]),$([H1][#6]),$(C([#6])[#6])]=[CX3;$([H2]),$([H1][#6]),$(C([#6])[#6])] Alkene
# sp2 C may be substituted only by C or H - 
# does not hit ketenes and allenes, nor enamines, enols and the like

[CX2]#[CX2] Alkyne
# non-carbon substituents (e.g. alkynol ethers) are rather rare, thus no further discrimination

[CX3]=[CX2]=[CX3] Allene


# I.2: One Carbon-Hetero Bond
# ---------------------------


# I.2.1 Alkyl Halogenides

[ClX1][CX4] Alkylchloride
# will also hit chloromethylethers and the like, but no chloroalkenes, -alkynes or -aromats 
# a more restrictive version can be obtained by modifying the Alcohol string.

[FX1][CX4] Alkylfluoride

[BrX1][CX4] Alkylbromide

[IX1][CX4] Alkyliodide


# I.2.2 Alcohols and Ethers

[OX2H][CX4;!$(C([OX2H])[O,S,#7,#15])] Alcohol
# nonspecific definition, no acetals, aminals, and the like

[OX2H][CX4H2;!$(C([OX2H])[O,S,#7,#15])] Primary_alcohol

[OX2H][CX4H;!$(C([OX2H])[O,S,#7,#15])] Secondary_alcohol

[OX2H][CX4D4;!$(C([OX2H])[O,S,#7,#15])] Tertiary_alcohol

[OX2]([CX4;!$(C([OX2])[O,S,#7,#15,F,Cl,Br,I])])[CX4;!$(C([OX2])[O,S,#7,#15])] Dialkylether
# no acetals and the like; no enolethers

[SX2]([CX4;!$(C([OX2])[O,S,#7,#15,F,Cl,Br,I])])[CX4;!$(C([OX2])[O,S,#7,#15])] Dialkylthioether
# no acetals and the like; no enolethers

[OX2](c)[CX4;!$(C([OX2])[O,S,#7,#15,F,Cl,Br,I])] Alkylarylether
# no acetals and the like; no enolethers

[c][OX2][c] Diarylether

[SX2](c)[CX4;!$(C([OX2])[O,S,#7,#15,F,Cl,Br,I])] Alkylarylthioether

[c][SX2][c] Diarylthioether

[O+;!$([O]~[!#6]);!$([S]*~[#7,#8,#15,#16])] Oxonium
# can't be aromatic, thus O and not #8

# I.2.3 Amines

[NX3+0,NX4+;!$([N]~[!#6]);!$([N]*~[#7,#8,#15,#16])] Amine
# hits all amines (prim/sec/tert/quart), including ammonium salts, also enamines, but not amides, imides, aminals, ...

# the following amines include also the protonated forms 

[NX3H2+0,NX4H3+;!$([N][!C]);!$([N]*~[#7,#8,#15,#16])] Primary_aliph_amine

[NX3H1+0,NX4H2+;!$([N][!C]);!$([N]*~[#7,#8,#15,#16])] Secondary_aliph_amine

[NX3H0+0,NX4H1+;!$([N][!C]);!$([N]*~[#7,#8,#15,#16])] Tertiary_aliph_amine

[NX4H0+;!$([N][!C]);!$([N]*~[#7,#8,#15,#16])] Quaternary_aliph_ammonium

[NX3H2+0,NX4H3+]c Primary_arom_amine

[NX3H1+0,NX4H2+;!$([N][!c]);!$([N]*~[#7,#8,#15,#16])] Secondary_arom_amine

[NX3H0+0,NX4H1+;!$([N][!c]);!$([N]*~[#7,#8,#15,#16])] Tertiary_arom_amine

[NX4H0+;!$([N][!c]);!$([N]*~[#7,#8,#15,#16])] Quaternary_arom_ammonium

[NX3H1+0,NX4H2+;$([N]([c])[C]);!$([N]*~[#7,#8,#15,#16])] Secondary_mixed_amine

[NX3H0+0,NX4H1+;$([N]([c])([C])[#6]);!$([N]*~[#7,#8,#15,#16])] Tertiary_mixed_amine

[NX4H0+;$([N]([c])([C])[#6][#6]);!$([N]*~[#7,#8,#15,#16])] Quaternary_mixed_ammonium

[N+;!$([N]~[!#6]);!$(N=*);!$([N]*~[#7,#8,#15,#16])] Ammonium
# only C and H substituents allowed. Quaternary or protonated amines
# NX4+ or Nv4+ is not recognized by Daylight's depictmatch if less than four C are present


# I.2.4 Others

[SX2H][CX4;!$(C([SX2H])~[O,S,#7,#15])] Alkylthiol

[SX2]([CX4;!$(C([SX2])[O,S,#7,#15,F,Cl,Br,I])])[CX4;!$(C([SX2])[O,S,#7,#15])] Dialkylthioether

[SX2](c)[CX4;!$(C([SX2])[O,S,#7,#15])] Alkylarylthioether

[SX2D2][SX2D2] Disulfide

[OX2H][CX4;!$(C([OX2H])[O,S,#7,#15,F,Cl,Br,I])][CX4;!$(C([N])[O,S,#7,#15])][NX3;!$(NC=[O,S,N])] 1,2-Aminoalcohol
# does not hit alpha-amino acids, enaminoalcohols, 1,2-aminoacetals, o-aminophenols, etc.

[OX2H][CX4;!$(C([OX2H])[O,S,#7,#15])][CX4;!$(C([OX2H])[O,S,#7,#15])][OX2H] 1,2-Diol
# does not hit alpha-hydroxy acids, enolalcohols, 1,2-hydroxyacetals, 1,2-diphenols, etc.

[OX2H][CX4;!$(C([OX2H])([OX2H])[O,S,#7,#15])][OX2H] 1,1-Diol

[OX2H][OX2] Hydroperoxide
#does not neccessarily have to be connected to a carbon atom, includes also hydrotrioxides

[OX2D2][OX2D2] Peroxo

[LiX1][#6,#14] Organolithium_compounds

[MgX2][#6,#14] Organomagnesium_compounds
# not restricted to Grignard compounds, also dialkyl Mg

[!#1;!#5;!#6;!#7;!#8;!#9;!#14;!#15;!#16;!#17;!#33;!#34;!#35;!#52;!#53;!#85]~[#6;!-] Organometallic_compounds
# very general, includes all metals covalently bound to carbon 


# I.3: Two Carbon-Hetero Bonds (Carbonyl and Derivatives)
# ----------------------------

# I.3.1 Double Bond to Hetero

[$([CX3H][#6]),$([CX3H2])]=[OX1] Aldehyde
# hits aldehydes including formaldehyde

[#6][CX3](=[OX1])[#6] Ketone
# does not include oxo-groups connected to a (hetero-) aromatic ring

[$([CX3H][#6]),$([CX3H2])]=[SX1] Thioaldehyde

[#6][CX3](=[SX1])[#6] Thioketone
# does not include thioxo-groups connected to a (hetero-) aromatic ring

[NX2;$([N][#6]),$([NH]);!$([N][CX3]=[#7,#8,#15,#16])]=[CX3;$([CH2]),$([CH][#6]),$([C]([#6])[#6])] Imine
# nitrogen is not part of an amidelike strukture, nor of an aromatic ring, but can be part of an aminal or similar

[NX3+;!$([N][!#6]);!$([N][CX3]=[#7,#8,#15,#16])] Immonium

[NX2](=[CX3;$([CH2]),$([CH][#6]),$([C]([#6])[#6])])[OX2H] Oxime

[NX2](=[CX3;$([CH2]),$([CH][#6]),$([C]([#6])[#6])])[OX2][#6;!$(C=[#7,#8])] Oximether
# ether, not ester or amide; does not hit isoxazole


# I.3.2. Two Single Bonds to Hetero

[OX2]([#6;!$(C=[O,S,N])])[CX4;!$(C(O)(O)[!#6])][OX2][#6;!$(C=[O,S,N])] Acetal
# does not hit hydroxy-methylesters, ketenacetals, hemiacetals, orthoesters, etc.

[OX2H][CX4;!$(C(O)(O)[!#6])][OX2][#6;!$(C=[O,S,N])] Hemiacetal

[NX3v3;!$(NC=[#7,#8,#15,#16])]([#6])[CX4;!$(C(N)(N)[!#6])][NX3v3;!$(NC=[#7,#8,#15,#16])][#6] Aminal
# Ns are not part of an amide or similar. v3 ist to exclude nitro and similar groups

[NX3v3;!$(NC=[#7,#8,#15,#16])]([#6])[CX4;!$(C(N)(N)[!#6])][OX2H] Hemiaminal

[SX2]([#6;!$(C=[O,S,N])])[CX4;!$(C(S)(S)[!#6])][SX2][#6;!$(C=[O,S,N])] Thioacetal

[SX2]([#6;!$(C=[O,S,N])])[CX4;!$(C(S)(S)[!#6])][OX2H] Thiohemiacetal

[NX3v3,SX2,OX2;!$(*C=[#7,#8,#15,#16])][CX4;!$(C([N,S,O])([N,S,O])[!#6])][FX1,ClX1,BrX1,IX1] Halogen_acetal_like
# hits chloromethylenethers and other reactive alkylating agents

[NX3v3,SX2,OX2;!$(*C=[#7,#8,#15,#16])][CX4;!$(C([N,S,O])([N,S,O])[!#6])][FX1,ClX1,BrX1,IX1,NX3v3,SX2,OX2;!$(*C=[#7,#8,#15,#16])] Acetal_like
# includes all of the above and other combinations (S-C-N, hydrates, ...), but still no aminomethylenesters and similar

[NX3v3,SX2,OX2;$(**=[#7,#8,#15,#16])][CX4;!$(C([N,S,O])([N,S,O])[!#6])][FX1,ClX1,BrX1,IX1] Halogenmethylen_ester_and_similar
# also reactive alkylating agents. Acid does not have to be carboxylic acid, also S- and P-based acids allowed

[NX3v3,SX2,OX2;$(**=[#7,#8,#15,#16])][CX4;!$(C([N,S,O])([N,S,O])[!#6])][NX3v3,SX2,OX2;!$(*C=[#7,#8,#15,#16])] NOS_methylen_ester_and_similar
# Same as above, but N,O or S instead of halogen. Ester/amide allowed only on one side 

[NX3v3,SX2,OX2;$(**=[#7,#8,#15,#16])][CX4;!$(C([N,S,O])([N,S,O])[!#6])][FX1,ClX1,BrX1,IX1,NX3v3,SX2,OX2;!$(*C=[#7,#8,#15,#16])] Hetero_methylen_ester_and_similar
# Combination of the last two patterns

[NX1]#[CX2][CX4;$([CH2]),$([CH]([CX2])[#6]),$(C([CX2])([#6])[#6])][OX2H] Cyanhydrine


# I.3.3 Single Bond to Hetero, C=C Double Bond (Enols and Similar)

[ClX1][CX3]=[CX3] Chloroalkene

[FX1][CX3]=[CX3] Fluoroalkene

[BrX1][CX3]=[CX3] Bromoalkene

[IX1][CX3]=[CX3] Iodoalkene

[OX2H][CX3;$([H1]),$(C[#6])]=[CX3] Enol
# no phenols

[OX2H][CX3;$([H1]),$(C[#6])]=[CX3;$([H1]),$(C[#6])][OX2H] Endiol
# no 1,2-diphenols, ketenacetals, ...

[OX2]([#6;!$(C=[N,O,S])])[CX3;$([H0][#6]),$([H1])]=[CX3] Enolether
# finds also endiodiethers, but not enolesters, no aromats

[OX2]([CX3]=[OX1])[#6X3;$([#6][#6]),$([H1])]=[#6X3;!$(C[OX2H])] Enolester


[NX3;$([NH2][CX3]),$([NH1]([CX3])[#6]),$([N]([CX3])([#6])[#6]);!$([N]*=[#7,#8,#15,#16])][CX3;$([CH]),$([C][#6])]=[CX3] Enamine
# does not hit amines attached to aromatic rings, nor may the nitrogen be aromatic

[SX2H][CX3;$([H1]),$(C[#6])]=[CX3] Thioenol

[SX2]([#6;!$(C=[N,O,S])])[CX3;$(C[#6]),$([CH])]=[CX3] Thioenolether


# I.4: Three Carbon-Hetero Bonds (Carboxyl and Derivatives)
# ------------------------------

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[ClX1] Acylchloride

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[FX1] Acylfluoride

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[BrX1] Acylbromide

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[IX1] Acyliodide

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[FX1,ClX1,BrX1,IX1] Acylhalide
# all of the above


# The following contains all simple carboxylic combinations of O, N, S, & Hal -
# - acids, esters, amides, ... as well as a few extra cases (anhydride, hydrazide...)
# Cyclic structures (including aromats) like lactones, lactames, ... got their own  
# definitions. Structures where both heteroatoms are part of an aromatic ring  
# (oxazoles, imidazoles, ...) were excluded.

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[$([OX2H]),$([OX1-])] Carboxylic_acid
# includes carboxylate anions

 Carboxylic_ester
# does not hit anhydrides or lactones

[#6][#6X3R](=[OX1])[#8X2][#6;!$(C=[O,N,S])] Lactone
# may also be aromatic

[CX3;$([H0][#6]),$([H1])](=[OX1])[#8X2][CX3;$([H0][#6]),$([H1])](=[OX1]) Carboxylic_anhydride
# anhydride formed by two carboxylic acids, no mixed anhydrides (e.g. between carboxylic acid and sulfuric acid); may be part of a ring, even aromatic

[$([#6X3H0][#6]),$([#6X3H])](=[!#6])[!#6] Carboxylic_acid_derivative
# includes most of the structures of I.4 and many more, also 1,3-heteroaromatics such as isoxazole

[CX3;!R;$([C][#6]),$([CH]);$([C](=[OX1])[$([SX2H]),$([SX1-])]),$([C](=[SX1])[$([OX2H]),$([OX1-])])] Carbothioic_acid
# hits both tautomeric forms, as well as anions

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[SX2][#6;!$(C=[O,N,S])] Carbothioic_S_ester

[#6][#6X3R](=[OX1])[#16X2][#6;!$(C=[O,N,S])] Carbothioic_S_lactone
# may also be aromatic

[CX3;$([H0][#6]),$([H1])](=[SX1])[OX2][#6;!$(C=[O,N,S])] Carbothioic_O_ester

[#6][#6X3R](=[SX1])[#8X2][#6;!$(C=[O,N,S])] Carbothioic_O_lactone

[CX3;$([H0][#6]),$([H1])](=[SX1])[FX1,ClX1,BrX1,IX1] Carbothioic_halide

[CX3;!R;$([C][#6]),$([CH]);$([C](=[SX1])[SX2H])] Carbodithioic_acid

[CX3;!R;$([C][#6]),$([CH]);$([C](=[SX1])[SX2][#6;!$(C=[O,N,S])])] Carbodithioic_ester

[#6][#6X3R](=[SX1])[#16X2][#6;!$(C=[O,N,S])] Carbodithiolactone


[CX3;$([R0][#6]),$([H1R0])](=[OX1])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Amide
# does not hit lactames

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[NX3H2] Primary_amide

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[#7X3H1][#6;!$(C=[O,N,S])] Secondary_amide

[CX3;$([R0][#6]),$([H1R0])](=[OX1])[#7X3H0]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])] Tertiary_amide

[#6R][#6X3R](=[OX1])[#7X3;$([H1][#6;!$(C=[O,N,S])]),$([H0]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Lactam
# cyclic amides, may also be aromatic

[#6X3;$([H0][#6]),$([H1])](=[OX1])[#7X3H0]([#6])[#6X3;$([H0][#6]),$([H1])](=[OX1]) Alkyl_imide
# may be part of a ring, even aromatic. only C allowed at central N. May also be triacyl amide

[#6X3;$([H0][#6]),$([H1])](=[OX1])[#7X3H0]([!#6])[#6X3;$([H0][#6]),$([H1])](=[OX1]) N_hetero_imide
# everything else than H or C at central N

[#6X3;$([H0][#6]),$([H1])](=[OX1])[#7X3H1][#6X3;$([H0][#6]),$([H1])](=[OX1]) Imide_acidic
# can be deprotonated

[$([CX3;!R][#6]),$([CX3H;!R])](=[SX1])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Thioamide
# does not hit thiolactames

[#6R][#6X3R](=[SX1])[#7X3;$([H1][#6;!$(C=[O,N,S])]),$([H0]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Thiolactam
# cyclic thioamides, may also be aromatic


[#6X3;$([H0][#6]),$([H1])](=[OX1])[#8X2][#7X2]=,:[#6X3;$([H0]([#6])[#6]),$([H1][#6]),$([H2])] Oximester
# may also be part of a ring / aromatic

[NX3;!$(NC=[O,S])][CX3;$([CH]),$([C][#6])]=[NX2;!$(NC=[O,S])] Amidine
# only basic amidines, not as part of aromatic ring (e.g. imidazole)

[CX3;$([H0][#6]),$([H1])](=[OX1])[#7X3;$([H1]),$([H0][#6;!$(C=[O,N,S])])][$([OX2H]),$([OX1-])] Hydroxamic_acid

[CX3;$([H0][#6]),$([H1])](=[OX1])[#7X3;$([H1]),$([H0][#6;!$(C=[O,N,S])])][OX2][#6;!$(C=[O,N,S])] Hydroxamic_acid_ester
# does not hit anhydrides of carboxylic acids withs hydroxamic acids


[CX3R0;$([H0][#6]),$([H1])](=[NX2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[$([OX2H]),$([OX1-])] Imidoacid
# not cyclic 

[#6R][#6X3R](=,:[#7X2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[$([OX2H]),$([OX1-])] Imidoacid_cyclic
# the enamide-form of lactames. may be aromatic like 2-hydroxypyridine 

[CX3R0;$([H0][#6]),$([H1])](=[NX2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[OX2][#6;!$(C=[O,N,S])] Imidoester
# esters of the above structures. no anhydrides. 

[#6R][#6X3R](=,:[#7X2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[OX2][#6;!$(C=[O,N,S])] Imidolactone
# no oxazoles and similar

[CX3R0;$([H0][#6]),$([H1])](=[NX2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[$([SX2H]),$([SX1-])] Imidothioacid
# not cyclic 

[#6R][#6X3R](=,:[#7X2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[$([SX2H]),$([SX1-])] Imidothioacid_cyclic
# the enamide-form of thiolactames. may be aromatic like 2-thiopyridine 

[CX3R0;$([H0][#6]),$([H1])](=[NX2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[SX2][#6;!$(C=[O,N,S])] Imidothioester
# thioesters of the above structures. no anhydrides. 

[#6R][#6X3R](=,:[#7X2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[SX2][#6;!$(C=[O,N,S])] Imidothiolactone
# no thioxazoles and similar

[#7X3v3;!$(N([#6X3]=[#7X2])C=[O,S])][CX3R0;$([H1]),$([H0][#6])]=[NX2v3;!$(N(=[#6X3][#7X3])C=[O,S])] Amidine
# only basic amidines, not substituted by carbonyl or thiocarbonyl, not as part of a ring 

[#6][#6X3R;$([H0](=[NX2;!$(N(=[#6X3][#7X3])C=[O,S])])[#7X3;!$(N([#6X3]=[#7X2])C=[O,S])]),$([H0](-[NX3;!$(N([#6X3]=[#7X2])C=[O,S])])=,:[#7X2;!$(N(=[#6X3][#7X3])C=[O,S])])] Imidolactam
# one of the two C~N bonds is part of a ring (may be aromatic), but not both - thus no imidazole

[CX3R0;$([H0][#6]),$([H1])](=[NX2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[FX1,ClX1,BrX1,IX1] Imidoylhalide
# not cyclic

[#6R][#6X3R](=,:[#7X2;$([H1]),$([H0][#6;!$(C=[O,N,S])])])[FX1,ClX1,BrX1,IX1] Imidoylhalide_cyclic
# may also be aromatic
 
# may be ring, aromatic, substituted with carbonyls, hetero, ...
# (everything else would get too complicated)

[$([$([#6X3][#6]),$([#6X3H])](=[#7X2v3])[#7X3v3][#7X3v3]),$([$([#6X3][#6]),$([#6X3H])]([#7X3v3])=[#7X2v3][#7X3v3])] Amidrazone
# hits both tautomers. as above, it may be ring, aromatic, substituted with carbonyls, hetero, ...


[NX3,NX4+;!$([N]~[!#6]);!$([N]*~[#7,#8,#15,#16])][C][CX3](=[OX1])[OX2H,OX1-] Alpha_aminoacid
# N may be alkylated, but not part of an amide (as in peptides), ionic forms are included
# includes also non-natural aminoacids with double-bonded or two aliph./arom. substituents at alpha-C 
# N may not be aromatic as in 1H-pyrrole-2-carboxylic acid

[OX2H][C][CX3](=[OX1])[OX2H,OX1-] Alpha_hydroxyacid

[NX3;$([N][CX3](=[OX1])[C][NX3,NX4+])][C][CX3](=[OX1])[NX3;$([N][C][CX3](=[OX1])[NX3,OX2,OX1-])] Peptide_middle
# finds peptidic structures which are neither C- nor N-terminal. Both neighbours must be amino-acids/peptides

[NX3;$([N][CX3](=[OX1])[C][NX3,NX4+])][C][CX3](=[OX1])[OX2H,OX1-] Peptide_C_term
# finds C-terminal amino acids

[NX3,NX4+;!$([N]~[!#6]);!$([N]*~[#7,#8,#15,#16])][C][CX3](=[OX1])[NX3;$([N][C][CX3](=[OX1])[NX3,OX2,OX1-])] Peptide_N_term
# finds N-terminal amino acids. As above, N may be substituted, but not part of an amide-bond.


[#6][OX2][CX4;$(C[#6]),$([CH])]([OX2][#6])[OX2][#6] Carboxylic_orthoester
# hits also anhydride like struktures (e. g. HC(OMe)2-OC=O residues)

[CX3]=[CX2]=[OX1] Ketene

[#7X2,#8X3,#16X2;$(*[#6,#14])][#6X3]([#7X2,#8X3,#16X2;$(*[#6,#14])])=[#6X3] Ketenacetal
# includes aminals, silylacetals, ketenesters, etc. C=C DB is not aromatic, everything else may be

[NX1]#[CX2] Nitrile
# includes cyanhydrines

[CX1-]#[NX2+] Isonitrile


[#6X3](=[OX1])[#6X3]=,:[#6X3][#7,#8,#16,F,Cl,Br,I] Vinylogous_carbonyl_or_carboxyl_derivative
# may be part of a ring, even aromatic

[#6X3](=[OX1])[#6X3]=,:[#6X3][$([OX2H]),$([OX1-])] Vinylogous_acid

[#6X3](=[OX1])[#6X3]=,:[#6X3][#6;!$(C=[O,N,S])] Vinylogous_ester

[#6X3](=[OX1])[#6X3]=,:[#6X3][#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Vinylogous_amide

[#6X3](=[OX1])[#6X3]=,:[#6X3][FX1,ClX1,BrX1,IX1] Vinylogous_halide



# I.5: Four Carbon-Hetero Bonds (Carbonic Acid and Derivatives)
# -----------------------------

[#6;!$(C=[O,N,S])][#8X2][#6X3](=[OX1])[#8X2][#6;!$(C=[O,N,S])] Carbonic_acid_dieester
# may be part of a ring, even aromatic

[#6;!$(C=[O,N,S])][OX2;!R][CX3](=[OX1])[OX2][FX1,ClX1,BrX1,IX1] Carbonic_acid_esterhalide

[#6;!$(C=[O,N,S])][OX2;!R][CX3](=[OX1])[$([OX2H]),$([OX1-])] Carbonic_acid_monoester
# unstable

[!#6][#6X3](=[!#6])[!#6] Carbonic_acid_derivatives


[#6;!$(C=[O,N,S])][#8X2][#6X3](=[SX1])[#8X2][#6;!$(C=[O,N,S])] Thiocarbonic_acid_dieester
# may be part of a ring, even aromatic

[#6;!$(C=[O,N,S])][OX2;!R][CX3](=[SX1])[OX2][FX1,ClX1,BrX1,IX1] Thiocarbonic_acid_esterhalide

[#6;!$(C=[O,N,S])][OX2;!R][CX3](=[SX1])[$([OX2H]),$([OX1-])] Thiocarbonic_acid_monoester


[#7X3;!$([#7][!#6])][#6X3](=[OX1])[#7X3;!$([#7][!#6])] Urea
# no check whether part of imide, biuret, etc. Aromatic structures are only hit if
# both N share no double bonds, like in the dioxo-form of uracil

[#7X3;!$([#7][!#6])][#6X3](=[SX1])[#7X3;!$([#7][!#6])] Thiourea

[#7X2;!$([#7][!#6])]=,:[#6X3]([#8X2&!$([#8][!#6]),OX1-])[#7X3;!$([#7][!#6])] Isourea
# O may be substituted. no check whether further amide-like bonds are present. Aromatic 
# structures are only hit if single bonded N shares no additional double bond, like in
# the 1-hydroxy-3-oxo form of uracil

[#7X2;!$([#7][!#6])]=,:[#6X3]([#16X2&!$([#16][!#6]),SX1-])[#7X3;!$([#7][!#6])] Isothiourea

[N;v3X3,v4X4+][CX3](=[N;v3X2,v4X3+])[N;v3X3,v4X4+] Guanidine
# also hits guanidinium salts. v3 and v4 to avoid nitroamidines

[NX3]C(=[OX1])[O;X2H,X1-] Carbaminic_acid
# quite unstable, unlikely to be found. Also hits salts

[#7X3][#6](=[OX1])[#8X2][#6] Urethan
# also hits when part of a ring, no check whether the last C is part of carbonyl

[#7X3][#6](=[OX1])[#7X3][#6](=[OX1])[#7X3] Biuret

[#7X3][#7X3][#6X3]([#7X3;!$([#7][#7])])=[OX1] Semicarbazide

[#7X3][#7X3][#6X3]([#7X3][#7X3])=[OX1] Carbazide

[#7X2](=[#6])[#7X3][#6X3]([#7X3;!$([#7][#7])])=[OX1] Semicarbazone

[#7X2](=[#6])[#7X3][#6X3]([#7X3][#7X3])=[OX1] Carbazone

[#7X3][#7X3][#6X3]([#7X3;!$([#7][#7])])=[SX1] Thiosemicarbazide

[#7X3][#7X3][#6X3]([#7X3][#7X3])=[SX1] Thiocarbazide

[#7X2](=[#6])[#7X3][#6X3]([#7X3;!$([#7][#7])])=[SX1] Thiosemicarbazone

[#7X2](=[#6])[#7X3][#6X3]([#7X3][#7X3])=[SX1] Thiocarbazone


[NX2]=[CX2]=[OX1] Isocyanate

[OX2][CX2]#[NX1] Cyanate

[NX2]=[CX2]=[SX1] Isothiocyanate

[SX2][CX2]#[NX1] Thiocyanate

[NX2]=[CX2]=[NX2] Carbodiimide

[CX4H0]([O,S,#7])([O,S,#7])([O,S,#7])[O,S,#7,F,Cl,Br,I] Orthocarbonic_derivatives
# halogen allowed just once, to avoid mapping to -OCF3 and similar groups (much more
# stable as for example C(OCH3)4)


# I.6 Aromatics
# -------------

# I know that this classification is not very logical, arylamines are found under I.2 ...

[OX2H][c] Phenol

[OX2H][c][c][OX2H] 1,2-Diphenol

[Cl][c] Arylchloride

[F][c] Arylfluoride

[Br][c] Arylbromide

[I][c] Aryliodide

[SX2H][c] Arylthiol

[c]=[NX2;$([H1]),$([H0][#6;!$([C]=[N,S,O])])] Iminoarene
# N may be substituted with H or C, but not carbonyl or similar
# aromatic atom is always C, not S or P (these are not planar when substituted)

[c]=[OX1] Oxoarene

[c]=[SX1] Thioarene

[nX3H1+0] Hetero_N_basic_H
# as in pyrole. uncharged to exclude pyridinium ions

[nX3H0+0] Hetero_N_basic_no_H
# as in N-methylpyrole. uncharged to exclude pyridinium ions

[nX2,nX3+] Hetero_N_nonbasic
# as in pyridine, pyridinium

[o] Hetero_O

[sX2] Hetero_S
# X2 because Daylight's depictmatch falsely describes C1=CS(=O)C=C1 as aromatic
# (is not planar because of lonepair at S)

[a;!c] Heteroaromatic


# Part II: N, S, P, Si, B 
# =======================


# II.1 Nitrogen
# -------------

[NX2](=[OX1])[O;$([X2]),$([X1-])] Nitrite
# hits nitrous acid, its anion, esters, and other O-substituted derivatives 

[SX2][NX2]=[OX1] Thionitrite

[$([NX3](=[OX1])(=[OX1])[O;$([X2]),$([X1-])]),$([NX3+]([OX1-])(=[OX1])[O;$([X2]),$([X1-])])] Nitrate
# hits nitric acid, its anion, esters, and other O-substituted derivatives 

[$([NX3](=O)=O),$([NX3+](=O)[O-])][!#8] Nitro
# hits nitro groups attached to C,N, ... but not nitrates

[NX2](=[OX1])[!#7;!#8] Nitroso
# no nitrites, no nitrosamines

[NX1]~[NX2]~[NX2,NX1] Azide
# hits both mesomeric forms, also anion

[CX3](=[OX1])[NX2]~[NX2]~[NX1] Acylazide

[$([#6]=[NX2+]=[NX1-]),$([#6-]-[NX2+]#[NX1])] Diazo

[#6][NX2+]#[NX1] Diazonium

[#7;!$(N*=O)][NX2]=[OX1] Nitrosamine

[NX2](=[OX1])N-*=O Nitrosamide
# includes nitrososulfonamides

[$([#7+][OX1-]),$([#7v5]=[OX1]);!$([#7](~[O])~[O]);!$([#7]=[#7])] N-Oxide
# Hits both forms. Won't hit azoxy, nitro, nitroso, or nitrate.


[NX3;$([H2]),$([H1][#6]),$([H0]([#6])[#6]);!$(NC=[O,N,S])][NX3;$([H2]),$([H1][#6]),$([H0]([#6])[#6]);!$(NC=[O,N,S])] Hydrazine
# no hydrazides

[NX3;$([H2]),$([H1][#6]),$([H0]([#6])[#6]);!$(NC=[O,N,S])][NX2]=[#6] Hydrazone

[NX3;$([H2]),$([H1][#6]),$([H0]([#6])[#6]);!$(NC=[O,N,S])][OX2;$([H1]),$(O[#6;!$(C=[N,O,S])])] Hydroxylamine
# no discrimination between O-, N-, and O,N-substitution


# II.2 Sulfur
# -----------

[$([SX4](=[OX1])(=[OX1])([#6])[#6]),$([SX4+2]([OX1-])([OX1-])([#6])[#6])] Sulfon
# can't be aromatic, thus S and not #16

[$([SX3](=[OX1])([#6])[#6]),$([SX3+]([OX1-])([#6])[#6])] Sulfoxide

[S+;!$([S]~[!#6]);!$([S]*~[#7,#8,#15,#16])] Sulfonium
# can't be aromatic, thus S and not #16

[SX4](=[OX1])(=[OX1])([$([OX2H]),$([OX1-])])[$([OX2H]),$([OX1-])] Sulfuric_acid
# includes anions

[SX4](=[OX1])(=[OX1])([$([OX2H]),$([OX1-])])[OX2][#6;!$(C=[O,N,S])] Sulfuric_monoester

[SX4](=[OX1])(=[OX1])([OX2][#6;!$(C=[O,N,S])])[OX2][#6;!$(C=[O,N,S])] Sulfuric_diester

[SX4](=[OX1])(=[OX1])([#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])])[$([OX2H]),$([OX1-])] Sulfuric_monoamide

[SX4](=[OX1])(=[OX1])([#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Sulfuric_diamide

[SX4](=[OX1])(=[OX1])([#7X3][#6;!$(C=[O,N,S])])[OX2][#6;!$(C=[O,N,S])] Sulfuric_esteramide

[SX4D4](=[!#6])(=[!#6])([!#6])[!#6] Sulfuric_derivative
# everything else (would not be a "true" derivative of sulfuric acid, if one of the substituents were less electronegative 
# than sulfur, but this should be very very rare, anyway)



#### sulfurous acid and derivatives missing!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!




[SX4;$([H1]),$([H0][#6])](=[OX1])(=[OX1])[$([OX2H]),$([OX1-])] Sulfonic_acid

[SX4;$([H1]),$([H0][#6])](=[OX1])(=[OX1])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Sulfonamide

[SX4;$([H1]),$([H0][#6])](=[OX1])(=[OX1])[OX2][#6;!$(C=[O,N,S])] Sulfonic_ester

[SX4;$([H1]),$([H0][#6])](=[OX1])(=[OX1])[FX1,ClX1,BrX1,IX1] Sulfonic_halide

[SX4;$([H1]),$([H0][#6])](=[!#6])(=[!#6])[!#6] Sulfonic_derivative
# includes all of the above and many more
# for comparison: this is what "all sulfonic derivatives but not the ones above" would look like:
# [$([SX4;$([H1]),$([H0][#6])](=[!#6])(=[!#6;!O])[!#6]),$([SX4;$([H1]),$([H0][#6])](=[OX1])(=[OX1])[!$([FX1,ClX1,BrX1,IX1]);!$([#6]);!$([OX2H]);!$([OX1-]);!$([OX2][#6;!$(C=[O,N,S])]);!$([#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])])])]


[SX3;$([H1]),$([H0][#6])](=[OX1])[$([OX2H]),$([OX1-])] Sulfinic_acid

[SX3;$([H1]),$([H0][#6])](=[OX1])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Sulfinic_amide

[SX3;$([H1]),$([H0][#6])](=[OX1])[OX2][#6;!$(C=[O,N,S])] Sulfinic_ester

[SX3;$([H1]),$([H0][#6])](=[OX1])[FX1,ClX1,BrX1,IX1] Sulfinic_halide

[SX3;$([H1]),$([H0][#6])](=[!#6])[!#6] Sulfinic_derivative

[SX2;$([H1]),$([H0][#6])][$([OX2H]),$([OX1-])] Sulfenic_acid

[SX2;$([H1]),$([H0][#6])][#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Sulfenic_amide

[SX2;$([H1]),$([H0][#6])][OX2][#6;!$(C=[O,N,S])] Sulfenic_ester

[SX2;$([H1]),$([H0][#6])][FX1,ClX1,BrX1,IX1] Sulfenic_halide

[SX2;$([H1]),$([H0][#6])][!#6] Sulfenic_derivative


# II.3 Phosphorous
# ----------------

[PX3;$([H3]),$([H2][#6]),$([H1]([#6])[#6]),$([H0]([#6])([#6])[#6])] Phosphine
# similar to amine, but less restrictive: includes also amide- and aminal-analogues 

[PX4;$([H3]=[OX1]),$([H2](=[OX1])[#6]),$([H1](=[OX1])([#6])[#6]),$([H0](=[OX1])([#6])([#6])[#6])] Phosphine_oxide

[P+;!$([P]~[!#6]);!$([P]*~[#7,#8,#15,#16])] Phosphonium
# similar to Ammonium

[PX4;$([H3]=[CX3]),$([H2](=[CX3])[#6]),$([H1](=[CX3])([#6])[#6]),$([H0](=[CX3])([#6])([#6])[#6])] Phosphorylen


# conventions for the following acids and derivatives: 
# acids find protonated and deprotonated acids
# esters do not find mixed anhydrides ( ...P-O-C(=O))
# derivatives: subtituents which go in place of the OH and =O are not H or C (may also be O, 
# thus including acids and esters)

[PX4;$([H1]),$([H0][#6])](=[OX1])([$([OX2H]),$([OX1-])])[$([OX2H]),$([OX1-])] Phosphonic_acid
# includes anions

[PX4;$([H1]),$([H0][#6])](=[OX1])([$([OX2H]),$([OX1-])])[OX2][#6;!$(C=[O,N,S])] Phosphonic_monoester

[PX4;$([H1]),$([H0][#6])](=[OX1])([OX2][#6;!$(C=[O,N,S])])[OX2][#6;!$(C=[O,N,S])] Phosphonic_diester

[PX4;$([H1]),$([H0][#6])](=[OX1])([$([OX2H]),$([OX1-])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphonic_monoamide

[PX4;$([H1]),$([H0][#6])](=[OX1])([#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphonic_diamide

[PX4;$([H1]),$([H0][#6])](=[OX1])([OX2][#6;!$(C=[O,N,S])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphonic_esteramide

[PX4;$([H1]),$([H0][#6])](=[!#6])([!#6])[!#6] Phosphonic_acid_derivative
# all of the above and much more 


[PX4D4](=[OX1])([$([OX2H]),$([OX1-])])([$([OX2H]),$([OX1-])])[$([OX2H]),$([OX1-])] Phosphoric_acid
# includes anions

[PX4D4](=[OX1])([$([OX2H]),$([OX1-])])([$([OX2H]),$([OX1-])])[OX2][#6;!$(C=[O,N,S])] Phosphoric_monoester

[PX4D4](=[OX1])([$([OX2H]),$([OX1-])])([OX2][#6;!$(C=[O,N,S])])[OX2][#6;!$(C=[O,N,S])] Phosphoric_diester

[PX4D4](=[OX1])([OX2][#6;!$(C=[O,N,S])])([OX2][#6;!$(C=[O,N,S])])[OX2][#6;!$(C=[O,N,S])] Phosphoric_triester

[PX4D4](=[OX1])([$([OX2H]),$([OX1-])])([$([OX2H]),$([OX1-])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphoric_monoamide

[PX4D4](=[OX1])([$([OX2H]),$([OX1-])])([#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphoric_diamide

[PX4D4](=[OX1])([#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])])([#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphoric_triamide

[PX4D4](=[OX1])([$([OX2H]),$([OX1-])])([OX2][#6;!$(C=[O,N,S])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphoric_monoestermonoamide

[PX4D4](=[OX1])([OX2][#6;!$(C=[O,N,S])])([OX2][#6;!$(C=[O,N,S])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphoric_diestermonoamide

[PX4D4](=[OX1])([OX2][#6;!$(C=[O,N,S])])([#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphoric_monoesterdiamide

[PX4D4](=[!#6])([!#6])([!#6])[!#6] Phosphoric_acid_derivative


[PX4;$([H2]),$([H1][#6]),$([H0]([#6])[#6])](=[OX1])[$([OX2H]),$([OX1-])] Phosphinic_acid

[PX4;$([H2]),$([H1][#6]),$([H0]([#6])[#6])](=[OX1])[OX2][#6;!$(C=[O,N,S])] Phosphinic_ester

[PX4;$([H2]),$([H1][#6]),$([H0]([#6])[#6])](=[OX1])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphinic_amide

[PX4;$([H2]),$([H1][#6]),$([H0]([#6])[#6])](=[!#6])[!#6] Phosphinic_acid_derivative


[PX3;$([H1]),$([H0][#6])]([$([OX2H]),$([OX1-])])[$([OX2H]),$([OX1-])] Phosphonous_acid

[PX3;$([H1]),$([H0][#6])]([$([OX2H]),$([OX1-])])[OX2][#6;!$(C=[O,N,S])] Phosphonous_monoester

[PX3;$([H1]),$([H0][#6])]([OX2][#6;!$(C=[O,N,S])])[OX2][#6;!$(C=[O,N,S])] Phosphonous_diester

[PX3;$([H1]),$([H0][#6])]([$([OX2H]),$([OX1-])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphonous_monoamide

[PX3;$([H1]),$([H0][#6])]([#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphonous_diamide

[PX3;$([H1]),$([H0][#6])]([OX2][#6;!$(C=[O,N,S])])[#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphonous_esteramide

[PX3;$([D2]),$([D3][#6])]([!#6])[!#6] Phosphonous_derivatives


[PX3;$([H2]),$([H1][#6]),$([H0]([#6])[#6])][$([OX2H]),$([OX1-])] Phosphinous_acid

[PX3;$([H2]),$([H1][#6]),$([H0]([#6])[#6])][OX2][#6;!$(C=[O,N,S])] Phosphinous_ester

[PX3;$([H2]),$([H1][#6]),$([H0]([#6])[#6])][#7X3;$([H2]),$([H1][#6;!$(C=[O,N,S])]),$([#7]([#6;!$(C=[O,N,S])])[#6;!$(C=[O,N,S])])] Phosphinous_amide

[PX3;$([H2]),$([H1][#6]),$([H0]([#6])[#6])][!#6] Phosphinous_derivatives


# II.4 Silicon
# ------------

[SiX4]([#6])([#6])([#6])[#6] Quart_silane
# four C-substituents. non-reactive, non-toxic, in experimental phase for drug development

[SiX4;$([H1]([#6])([#6])[#6]),$([H2]([#6])[#6]),$([H3][#6]),$([H4])] Non-quart_silane
# has 1-4 hydride(s), reactive. Daylight's depictmatch does not add hydrogens automatically to 
# the free positions at Si, thus Hs had to be added implicitly

[SiX4]([FX1,ClX1,BrX1,IX1])([#6])([#6])[#6] Silylmonohalide
# reagents for inserting protection groups

[SiX4]([!#6])([#6])([#6])[#6] Het_trialkylsilane
# mostly acid-labile protection groups such as trimethylsilyl-ethers 

[SiX4]([!#6])([!#6])([#6])[#6] Dihet_dialkylsilane

[SiX4]([!#6])([!#6])([!#6])[#6] Trihet_alkylsilane

[SiX4]([!#6])([!#6])([!#6])[!#6] Silicic_acid_derivative
# four substituent which are neither C nor H


# II.5 Boron
# ----------

[BX3]([#6])([#6])[#6] Trialkylborane
# also carbonyls allowed

[BX3]([!#6])([!#6])[!#6] Boric_acid_derivatives
# includes acids, esters, amides, ... H-substituent at B is very rare.

[BX3]([!#6])([!#6])[!#6] Boronic_acid_derivative
# # includes acids, esters, amides, ...

[BH1,BH2,BH3,BH4] Borohydride
# at least one H attached to B

[BX4] Quaternary_boron
# mostly borates (negative charge), in complex with Lewis-base



# Part III: Some Special Patterns
# ===============================


# III.1 Chains
# ------------

# some simple chains



# III.2 Rings
# -----------

a Aromatic

[!#6;!R0] Heterocyclic
# may be aromatic or not

[OX2r3]1[#6r3][#6r3]1 Epoxide
# toxic/reactive. may be annelated to aromat, but must not be aromatic itself (oxirane-2,3-dione)

[NX3H1r3]1[#6r3][#6r3]1 NH_aziridine
# toxic/reactive according to Maybridge's garbage filter

[D4R;$(*(@*)(@*)(@*)@*)] Spiro
# at least two different rings can be found which are sharing just one atom.
# these two rings can be connected by a third ring, so it matches also some 
# bridged systems, like morphine

[R;$(*(@*)(@*)@*);!$([R2;$(*(@*)(@*)(@*)@*)])]@[R;$(*(@*)(@*)@*);!$([R2;$(*(@*)(@*)(@*)@*)])] Annelated_rings
# two different rings sharing exactly two atoms

[R;$(*(@*)(@*)@*);!$([D4R;$(*(@*)(@*)(@*)@*)]);!$([R;$(*(@*)(@*)@*);!$([R2;$(*(@*)(@*)(@*)@*)])]@[R;$(*(@*)(@*)@*);!$([R2;$(*(@*)(@*)(@*)@*)])])] Bridged_rings
# part of two or more rings, not spiro, not annelated -> finds bridgehead atoms, 
# but only if they are not annelated at the same time - otherwise impossible (?)
# to distinguish from non-bridgehead annelated atoms

# some basic ring-patterns (just size, no other information):





# III.3 Sugars and Nucleosides/Nucleotides, Steroids
# --------------------------------------------------

# because of the large variety of sugar derivatives, different patterns can be applied.
# The choice of patterns and their combinations will depend on the contents of the database
# e.g. natural products, nucleoside analoges with modified sugars, ... as well as on the 
# desired restriction


[OX2;$([r5]1@C@C@C(O)@C1),$([r6]1@C@C@C(O)@C(O)@C1)] Sugar_pattern_1
# 5 or 6-membered ring containing one O and at least one (r5) or two (r6) oxygen-substituents. 

[OX2;$([r5]1@C(!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C1),$([r6]1@C(!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C@C1)] Sugar_pattern_2
# 5 or 6-membered ring containing one O and an acetal-like bond at postion 2. 

[OX2;$([r5]1@C(!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C(O)@C1),$([r6]1@C(!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C(O)@C(O)@C1)] Sugar_pattern_combi
# combination of the two above

[OX2;$([r5]1@C(!@[OX2H1])@C@C@C1),$([r6]1@C(!@[OX2H1])@C@C@C@C1)] Sugar_pattern_2_reducing
# 5 or 6-membered cyclic hemi-acetal

[OX2;$([r5]1@[C@@](!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C1),$([r6]1@[C@@](!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C@C1)] Sugar_pattern_2_alpha
# 5 or 6-membered cyclic hemi-acetal

[OX2;$([r5]1@[C@](!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C1),$([r6]1@[C@](!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C@C1)] Sugar_pattern_2_beta
# 5 or 6-membered cyclic hemi-acetal

##Poly_sugar_1: ([OX2;$([r5]1@C@C@C(O)@C1),$([r6]1@C@C@C(O)@C(O)@C1)].[OX2;$([r5]1@C@C@C(O)@C1),$([r6]1@C@C@C(O)@C(O)@C1)])
# pattern1 occours more than once (in same molecule, but moieties don't have to be adjacent!)

##Poly_sugar_2: ([OX2;$([r5]1@C(!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C1),$([r6]1@C(!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C@C1)].[OX2;$([r5]1@C(!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C1),$([r6]1@C(!@[OX2,NX3,SX2,FX1,ClX1,BrX1,IX1])@C@C@C@C1)])
# pattern2 occours more than once (in same molecule, but moieties don't have to be adjacent!)


# III.4 Everything else...
# ------------------------

*=*[*]=,#,:[*] Conjugated_double_bond

*#*[*]=,#,:[*] Conjugated_tripple_bond

*/[D2]=[D2]\* Cis_double_bond
# only one single-bonded substituent on each DB-atom. no aromats. 
# only found when character of DB is explicitely stated.

*/[D2]=[D2]/* Trans_double_bond
# analog

[$(*=O),$([#16,#14,#5]),$([#7]([#6]=[OX1]))][#8X2][$(*=O),$([#16,#14,#5]),$([#7]([#6]=[OX1]))] Mixed_anhydrides
# should hits all combinations of two acids

[FX1,ClX1,BrX1,IX1][!#6] Halogen_on_hetero

[F,Cl,Br,I;!$([X1]);!$([X0-])] Halogen_multi_subst
# Halogen which is not mono-substituted nor an anion, e.g. chlorate.
# Most of these cases should be also filtered by Halogen_on_hetero.

[FX1][CX4;!$([H0][Cl,Br,I]);!$([F][C]([F])([F])[F])]([FX1])([FX1]) Trifluoromethyl
# C with three F attached, connected to anything which is not another halogen

[#6]~[#7,#8,#16] C_ONS_bond
# probably all drug-like molecules have at least one O, N, or S connected to a C -> nice filter

## Mixture: (*).(*)
# two or more seperate parts, may also be salt
# component-level grouping is not yet supported in Open Babel Version 2.0


[!+0] Charged

[-1,-2,-3,-4,-5,-6,-7] Anion

[+1,+2,+3,+4,+5,+6,+7] Kation

([-1,-2,-3,-4,-5,-6,-7]).([+1,+2,+3,+4,+5,+6,+7]) Salt
# two or more seperate components with opposite charges

##Zwitterion: ([-1,-2,-3,-4,-5,-6,-7].[+1,+2,+3,+4,+5,+6,+7])
# both negative and positive charges somewhere within the same molecule. 

[$([#7X2,OX1,SX1]=*[!H0;!$([a;!n])]),$([#7X3,OX2,SX2;!H0]*=*),$([#7X3,OX2,SX2;!H0]*:n)] 1,3-Tautomerizable
# 1,3 migration of H allowed. Includes keto/enol and amide/enamide. 
# Aromatic rings must stay aromatic - no keto form of phenol 

[$([#7X2,OX1,SX1]=,:**=,:*[!H0;!$([a;!n])]),$([#7X3,OX2,SX2;!H0]*=**=*),$([#7X3,OX2,SX2;!H0]*=,:**:n)] 1,5-Tautomerizable

[!$(*#*)&!D1]-!@[!$(*#*)&!D1] Rotatable_bond
# taken from http://www.daylight.com/support/contrib/smarts/content.html

[CX3]=[CX3][$([CX3]=[O,N,S]),$(C#[N]),$([S,P]=[OX1]),$([NX3]=O),$([NX3+](=O)[O-])] Michael_acceptor
# the classical case: C=C near carbonyl, nitrile, nitro, or similar
# Oxo-heteroaromats and similar are not included.

[CX3](=[OX1])[NX2]=[NX2][CX3](=[OX1]) Dicarbodiazene
# Michael-like acceptor, see Mitsunobu reaction

# H-Bond_donor:

# H-Bond_acceptor:

# Pos_ionizable:

# Neg_ionizable:

# Unlikely_ions: 
# O+,N-,C+,C-, ...

[$([CX4;!$([H0]);!$(C[!#6;!$([P,S]=O);!$(N(~O)~O)])][$([CX3]=[O,N,S]),$(C#[N]),$([S,P]=[OX1]),$([NX3]=O),$([NX3+](=O)[O-]);!$(*[S,O,N;H1,H2]);!$([*+0][S,O;X1-])]),$([CX4;!$([H0])]1[CX3]=[CX3][CX3]=[CX3]1)] CH-acidic
# C-H alpha to carbony, nitro or similar, C is not double-bonded, only C, H, S,P=O and nitro substituents allowed. 
# pentadiene is included. acids, their salts, prim./sec. amides, and imides are excluded. 
# hits also CH-acidic_strong

[CX4;!$([H0]);!$(C[!#6;!$([P,S]=O);!$(N(~O)~O)])]([$([CX3]=[O,N,S]),$(C#[N]),$([S,P]=[OX1]),$([NX3]=O),$([NX3+](=O)[O-]);!$(*[S,O,N;H1,H2]);!$([*+0][S,O;X1-])])[$([CX3]=[O,N,S]),$(C#[N]),$([S,P]=[OX1]),$([NX3]=O),$([NX3+](=O)[O-]);!$(*[S,O,N;H1,H2]);!$([*+0][S,O;X1-])] CH-acidic_strong
# same as above (without pentadiene), but carbonyl or similar on two or three sides

[$([*@](~*)(~*)(*)*),$([*@H](*)(*)*),$([*@](~*)(*)*),$([*@H](~*)~*)] Chiral_center_specified
# Hits atoms with tetrahedral chirality, if chiral center is specified in the SMILES string
# depictmach does not find oxonium, sulfonium, or sulfoxides!

# Chiral_center_unspecified: [$([*@?](~*)(~*)(*)*),$([*@?H](*)(*)*),$([*@?](~*)(*)*),$([*@?H](~*)~*)]
# Hits atoms with tetrahedral chirality, if chiral center is not specified in the SMILES string
# "@?" (unspecified chirality) is not yet supported in Open Babel Version 2.0