Report:MARPAT on STN/Chemical Structure Tools/Assigning Match Levels

Generic Structure Searching: Assigning Match Levels

Every node in a MARPAT structure has a Markush attribute called a “match level.” Users may individually set match levels for nodes, or rely on the system defaults, which differ depending on the identity of the node. Changing default match levels can have a broadening or narrowing effect on the generic structure query.

Setting the “match level” on a node is done via this system window. The options are “Atom,” “Class,” or “Any.”

Match level “atom” is the most restrictive match level. Generic nodes in the query match only specific atoms in candidate answers. The following table from MARPAT search training materials illustrates the retrieval potential when a given node is designated match level “Atom.”

In sum, the specific atom (in this case the halogen atom Cl) will only retrieve an exact match, and a variable group (in this case, the variable group X, which stands for any halogen) will retrieve any specific atom match, but will NOT retrieve a candidate answer that has the same variable group at the given position. Using “Atom” match level on a variable group is not an inclusive strategy for finding all possible structure hits, as it will not pick up generically disclosed structure groups at that node, possibly excluding some Markush answers.

Match level “Class” is more inclusive: it allows specific atoms or generic nodes in the query to retrieve specific atoms and generic nodes in candidate answers. The following table from MARPAT search training materials illustrates the retrieval potential when a given node is designated match level “Class.”

As shown in this table, match level “Class” is inclusive of generically disclosed Markush structures.

Match level “any” also allows R-nodes, which are indefinite substituents requiring text notes, such as “protecting group” “dye” or “cation.” Because it is so inclusive, match level “any” should be used sparingly and only when appropriate for the search.

As shown in this table, match level “Any” is inclusive of generically disclosed Markush structures, and R-nodes. These R-nodes are structures with indefinite substituents in the file structure that are described with text terms. Text terms in MARPAT indexing may be used to specify a potential Markush structure group that is too generic even to be described using defined variable groups (discussed above). For example, a patent may specify that a structure has a vague attachment such as an “organic group”, “protecting group”, “anion”, or “group to form ring”. These structures fall outside the available MARPAT indexing tools used by CAS staff, and so they are added to the record as text notes. Using the match level “any” designation on a particular node will bring back structure hits having any text note at that position, as well as structure hits fulfilling the other possible criteria denoted by the node. Obviously, this match level designation is extremely inclusive; it should be used cautiously to prevent an overabundance of search hits.

Match level defaults are set logically; therefore they will rarely need to be changed. All atoms in rings and variables representing ring systems are assigned match level “atom,” by default. All other nodes are match level “class” by default. Match level “Any” is never the system default for any node. The most common situations where match levels should be changed are: 1) when the search is for a carbon chain of a defined length and the user has drawn the carbons rather than using Ak and restricting its element count, those carbon atoms should be changed to match level “atom”; and 2) when the search is for ring systems or cyclic variables and the user would like to find cyclic variables in the answer, rather than just defined ring systems, the match level should be changed to class.