Understanding what Proteins talk!

Suppose Proteins were a clan speaking a particular language what would they be? Some interesting questions would be 

• What would be their vocabulary, syllables, consonants, vowels ...?
• What would be their set of ‘allowed’ or ‘forbidden’ rules in forming their words or building units? 
• Could α-helices and ß-pleats be regarded as alphabets/words with ‘turns’ and ‘loops’ being vowels/conjunctions?
• Would ‘structural motifs’ form the words with ‘Domains’ forming the sentence/paragraph?

Some more applied and directly relevant questions could be

• Can we predict the function of a protein given its sequence?

• Can active site of a protein be modified with ‘Synonymns’ – equivalent amino acids with similar side group structure that doesn’t change the ‘meaning’ or function of the molecule?

            While ‘Associative learning’ would mean understanding to use sentences in context (as a small child does), ‘Grammatical learning’ would constitute to understanding the language structure. Present day bioinformatics is much akin to associative learning e.g. use of databases, pattern matching algorithms to understand the structure/function relationships. Some of the linguistic approaches include comparing two documents that have protein tags and print the results with precision, recall and f-scores (Click for more details). According to the recommendations of the Workshop on Language modeling of Biological data (2001), Some of the areas included

• Identifying the practical utility of non-stochastic grammars in key issues underlying macromolecular structure (including RNA secondary structures)

• Development of Grammars with structured primitives and their relevance to topological structures in CB.