BLAST & generative breakthroughs

The amount of flashy news on the Internet truly has to be studied. From eye-catching (and mind-disrupting) short form reels and TikToks to TV shows, movies, and video games with loud voices, bright colors, and exciting animations, much of technology has become about the complex, flashy components of life. Interestingly, in the world of biological data, pretty similar stuff is going on.

Computational biology enthusiasts (especially more recent ones: high schoolers like myself, or people who have recently gained an interest for the field) often refer to the profound CRISPR & AlphaFold algorithms as the versatile backbones of modern genomics. However, the impact of these sort of algorithms pales in comparison to that of the quiet but profound workhorse: BLAST.

BLAST, or the Basic Local Alignment Search Tool, was introduced in 1990 by Stephen Altschul and colleagues at the National Center for Biotechnology Information (NCBI).

It may not have the sci-fi allure of CRISPR or AlphaFold (the word "basic" is in its name), but BLAST is one of the most fundmentally critical algorithms to bioengineering.

Without it, tools like CRISPR may have never been properly invented.

By the late 1980s, labs had begun sequencing genes and proteins at a growing pace, but comparing these sequences of nucleotides (ATGC) to other sequences, to find similarities and differences, was extremely slow. Biologists a faster way to search through the exponentially-growing databased of genetic info. They had to rapidly answer questions like:

• “Does this gene exist in another species?”

• “Is this protein evolutionarily conserved?”

• “What might this sequence even do?”

Sure, alignment methods existed, such as the Smith-Waterman algorithm, but they weren't fast enough for databased with millions of sequences. A faster method was urgently needed.

Then came BLAST, a fast, heuristic-based algorithm that finds regions of local similarity between biological sequences. Unlike full-length global alignments (which again, are accurate but take too long), BLAST focused on high-scoring segments: short regions where sequences matched almost exactly. This was an excellent example of the 80/20 rule. BLAST didn't search the full gene and spend time on it, but instead, focused on critical segments that were biologically meaningful.

Its precise scope of focus made it extremely helpful. BLAST traded a small amount of accuracy for order-of-magnitude of improvement in terms of speed. And even though it was fast, it was still comparable to full-length algorithms. More importantly, BLAST made genes and sequencing far more user-friendly and accessible, breaking barriers in the otherwise complex field of computational biology. The pipeline of BLAST, while simple, had a really incredible impact on the efficiency of scientists.

BLAST basic pipeline:

1. Break query sequence into short “words.”

2. Search for these words in the database.

3. Extend matches in both directions.

4. Score and rank the alignments.

And most importantly, BLAST became infrastructure for a future generation of computational databases. GenBank, UniProt, UCSC Genome Browser, and more were created entirely based off of BLAST. The algorithm even inspired entire algorithm families: PSI-BLAST (position-specific), MegaBLAST (for longer sequences), and BLASTX, TBLASTN, and TBLASTX for translating between nucleotide and protein sequences.

Of course, BLAST isn’t perfect. It doesn’t perform well on very long or repetitive sequences, it isn’t the best for structural comparisons, and it’s been supplemented by faster algorithms like DIAMOND or MMseqs2 in big-data applications.

But its design is fast, user-centric, and way ahead of its time. remains iconic. Even today, 35 years later, biologists routinely paste a sequence into NCBI’s BLAST interface to answer real-world questions.

BLAST was the first time computers helped us see biological as a connected body of information. It turned bioinformatics from a hard-to-understand niche to an accessible yet still important necessity. And in doing so, BLAST powered the genomics revolution.

For a piece of code written in 1990, that’s a legacy few inventions can rival.