[dekhn]

Do what math now? I spent 10 years as an academic doing bioinformatics, with deep expertise in BLAST (I helped demonstrate that BLAST E-values accurately represent homology).

If you look, he shows some pages with resuults where the E-value is 282. Normally, an evalue should be 1e-2 or better (ideally 1e-6) before you start making ANY claims.

I've seen hundreds of examples where a person did the math, convinced everybody they were right, and either subtle errors or mistaken assumptions meant the work was useless. These days, if somebody starts with "Here, I'll show you how to use blast to demonstrate that the lab leak hypothesis is true", my priors say "most likely just garbage".

[flobosg]

The amount of people that use BLAST every day and cannot define E-values (despite being explained in its FAQ! [1]) is baffling.

[1]: https://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web&PAGE_TYPE=B...

[dekhn]

Yep! In my work we had a golden labelled set of known-homologous proteins, as well as sets of unknown-relationship and known-not-to-be-homologous (based on superfamily). We blasted all-vs-all and after dealing with a number of BLAST issues finally demonstrated that the searches produce false positives at the expected rates. I gained a lot of respect for the folks who implemented the statistical framework behind blast E-values.

I had another advisor who thought they were good at homology, so they'd take protein matches with an E-value of 10 (almost certainly random chance) and munge the data to make up a story. Most of my work was about using very stringent blast searches to make profiles, but all that work is superseded by HMMER now.

[throwawaymaths]

But this is not a homology exercise with natural selection. I can show you where I did a blast search, panned for "select amino acid substitutions of interest" and from the library out popped out a more efficient enzyme.

[dekhn]

Sure, show us.