Table 1. Sequence search algorithms and methods used for specificity assessments of oligonucleotides.

Time of introduction	Method	Advantages	Disadvantages
1990	BLAST individual drug candidates on a webpage	Can be performed by anyone. No programming skills or specialist software packages needed	Very low throughput, interpretation subjective, uses a heuristic algorithm, i.e. it is not exhaustive
2005	Scripted BLAST or FASTA on local databases	Can be performed on many RNA targets in a standardized fashion	BLAST and FASTA are designed and optimized to detect evolutionary relationships, not hybridization
2007	Suffix-array matching and Burrows–Wheeler Transforms	Fast and exhaustive. Can be performed for all possible gapmers against a target	Character matching is unphysical, large memory requirement
2010	Search with energy based affinity model	Thermodynamic scoring based on binding affinity between gapmer and RNA targets	Slow, parameters not publicly available for most chemical modifications
Future (nothing published yet)	Activity model trained on transcriptomics data and historic screening results	Driven by actual activity measurements on unintended RNA targets	Requires expertise and high-quality, needs comprehensive data to build

The times of introduction represent estimates based on our own experience, and lag between 1 and 10 years behind the time of publication in scientific journals. The methods listed are referenced in the main text when mentioned.