Subset of SMILES

[rapodaca]

The ms makes several references to "dialect" in the linguistic sense and this is of course the code name for the language. But the goal is not to make yet another dialect. The goal is to for the first time fully define a language that functions as a subset of SMILES-as-practiced. No extensions. No pet nice-to-haves. But a subset to the extent it's possible without internal inconsistencies.

Parts to improve:

0. Title
1. Line 37. The introduction leads to this point, so this paragraph must concisely spell out the aim. It doesn't quite hit the mark. It uses the "dialect" idea, for one.
2. Discussion section. Line 497 brings in SMILES. This is a good place to re-introduce the previous two papers on SMILES. Compare what's in them (and not in them) with Dialect.

[rapodaca]

Points of incompatibility. These are features explicitly defined in one or the other paper, but which are different in Dialect:

• new element symbols: Db; Sg; Bh; Hs; Mt; Ds; Rg; Cn; Nh; Fl; Mc; Lv; Ts; Og
• disallowed element symbols: Ha
• disallowed symbol , (JCICS)
• disallowed charges: --; ++; +++; etc.
• disallowed bonds: :
• double branching is disallowed, e.g.: *((*))*; *(((*)))*; etc.
• reactions are disallowed, e.g.: *>>*
• all stereo descriptors except @ and @@ are disallowed
  • e.g., @@@ and @@@@ (note recursion in grammar)
  • @1
  • @AL2, @AL2, @SP1, etc.
• hydrogen is just another bracket atom
  • [HH] is valid as are [HH2] and other chemically nonsensical constructs. Contradicts JCICS.
• . is NOT a "bond of formal order zero" (p 88, bottom). It is not a bond at all. . This point is contradicted elsewhere, so must be clarified.
• integer values restrictions (previously no upper bound)
  • isotope: 0 < value < 1000
  • charge: 0 < value < 10
  • hcount: 0 < value < 10
  • map: 0 <= value < 999

Points not explained by either paper, but in Dialect:

• computing implicit hydrogen
  • on a selected atom
• selection algorithm
• deselection algorithm
  • pruning
• LL(1) formal grammar

Contradictory points resolved in Dialect:

• table of default valences
  • 1988: N(3)
  • 2003: N(3,5)
• membership of organic subset
  • H is NOT in the "aromatic subset." See Hn1cccc1, JCICS p 35 bottom left, contradicts p32 top right.
• atom selectability
  • 1988: implies "aromatic cycles" only, although in a contradictory way
  • 2003: examples of acyclic selections

[rapodaca]

Table of differences:

Feature	SMILES	Dialect
element symbol Ha	accepts	rejects
element symbols Db; Sg; Bh; Hs; Mt; Rg; Cn; Nh; Fl; Mc; Lv; Ts; and Og	rejects	accepts
future element symbols approved by IUPAC	rejects	accepts
comma symbol (,)	may accept	rejects
multiple branching e.g., *((*))*)	accepts	rejects
reactions using greater than symbol (>)	accepts	rejects
extended stereodescriptors e.g., @@@, @@@@, @AL1, @1, and @SP1	accepts	rejects
use of stereodescriptors on odd cumulene centers	accepts	rejects
virtual hydrogen count on hydrogen	probably rejects	accepts
detachments are bonds of "formal order zero"	probably accepts	rejects
upper and lower bounds on atomic properties	rejects	accepts
nitrogen default valence includes 5	partially accepts	accepts
unbracketed hydrogen atom	partially accepts	rejects
acyclic atom selection	partially accepts	accepts