Some calculations suggest it is entirely feasible that for all practical (i.e. functional and structural) purposes, protein sequence space has been fully explored during the course of evolution of life on Earth.
[13] The starting gene can be mutagenised by random point mutations (by chemical mutagens or error prone PCR)[14][15] and insertions and deletions (by transposons).
Finally, specific regions of a gene can be systematically randomised[19] for a more focused approach based on structure and function knowledge.
[20] Therefore, a high-throughput assay is vital for measuring activity to find the rare variants with beneficial mutations that improve the desired properties.
[21][22] During in vivo evolution, each cell (usually bacteria or yeast) is transformed with a plasmid containing a different member of the variant library.
This format has the advantage of selecting for properties in a cellular environment, which is useful when the evolved protein or RNA is to be used in living organisms.
When performed without cells, DE involves using in vitro transcription translation to produce proteins or RNA free in solution or compartmentalised in artificial microdroplets.
They are also less expensive and labour-intensive than screening, however they are typically difficult to engineer, prone to artefacts and give no information on the range of activities present in the library.
Each variant gene is individually expressed and assayed to quantitatively measure the activity (most often by a colourgenic or fluorogenic product).
Even the most high throughput assays usually have lower coverage than selection methods but give the advantage of producing detailed information on each one of the screened variants.
[30] A restriction of directed evolution is that a high-throughput assay is required in order to measure the effects of a large number of different random mutations.
[33] Recent theoretical approaches have aimed to overcome the limitation of speed through an application of counter-diabatic driving techniques from statistical physics, though this has yet to be implemented in a directed evolution experiment.
[1][35] Beneficial mutations are rare, so large numbers of random mutants have to be screened to find improved variants.
'Focused libraries' concentrate on randomising regions thought to be richer in beneficial mutations for the mutagenesis step of DE.
[41] As a protein engineering tool, DE has been most successful in three areas: The study of natural evolution is traditionally based on extant organisms and their genes.
This allows for detailed measurements of evolutionary processes, for example epistasis, evolvability, adaptive constraint[60][61] fitness landscapes,[62] and neutral networks.
For example, global proteome-wide substitutions of natural amino acids with fluorinated analogs have been attempted in Escherichia coli[64] and Bacillus subtilis.
[65] A complete tryptophan substitution with thienopyrrole-alanine in response to 20899 UGG codons in Escherichia coli was reported in 2015 by Budisa and Söll.