In response to user requests, a Standard InChI (i.e. without options for properties such as tautomerism and stereoconfiguration) has been defined as follows:
- Standard InChI is for the purposes of interoperability/compatibility between large databases/web searching and information exchange.
- Standard InChI and non-standard InChI are always distinguishable.
- Standard InChI is a stable identifier; however, periodic updates may be necessary; they are reflected in the identifier version designation, which is included in the InChI string.
- Any shortcomings in standard InChI may be addressed using non-standard InChI (currently obtainable using InChI version 1.02beta).
In response to user feedback the format of InChIKey has been changed; it is different from that in InChI software v. 1.02-beta, having 27 characters rather than 25.
Standard InChIKey has five distinct components.
- 14-character hash of the basic (Mobile-H) InChI layer;
- 8-character hash of the remaining layers (except for the “/p” segment, which accounts for added or removed protons: it is not hashed at all; the number of protons is encoded at the end of the standard InChIKey.)
- 1 flag character,
- 1 version character
- the last character is a [de]protonation indicator.
The overall length of InChIKey is fixed at 27 characters, including separators (dashes):
This is significantly shorter than a typical InChI string.
(1) AAAAAAAAAAAAAA is a 14-character hash.
(2) BBBBBBBB is an 8-character hash
(3) F is a flag indicating standard InChIKey (produced out of standard InChI): it always has the value ‘S’.
(4) V is a flag for InChI version character: ‘A’ for version 1, ‘B’ for version 2, etc.
P is an indicator for the number of protons; this number is not encoded in the hash but is indicated as a separate 2-character block at the end, where one character is a hyphen, as –N for neutral, -M for -1 hydrogen, -O for +1 hydrogen, etc.
Full details and examples are provided in the documentation accompanying the software download.
Software implementing the final InChI version 1.02 for non-standard InChI (i.e. with all previous options retained and with the 27-character InChIKey) will be issued in due course.
Users are encouraged to report their experiences and any problems via the SourceForge website (http://sourceforge.net/projects/inchi).
Steve Heller (Chair, IUPAC InChI Subcommittee)
Alan McNaught (Coordinator, IUPAC InChI project)