iiinova — where your digital self keeps evolving

Closer view of that top branch and the nearby sequences. The "USA/FL-PRM-GKISBBB" one is the travel sequence from Cuba.

0100

These don't appear to be the result of a single introduction to New Brunswick. Bottom 4 seqs sprout from the major PJ.2.1 branch, but the top 3 descend from a branch with seqs from California, Colorado, and New Jersey—and are right next door to a US travel sequence from Cuba.

182

Reminder: PJ.2.1 is a saltation variant that originated in Ontario, Canada. Mutations in the branches leading up to it, starting with BA.2.86, pictured below.

1101

Ryan Hisner @ryanhisner.bsky.social · 18/09/2026

One of two antibodies that PJ.2.1 is more vulnerable to than any other (IC50 = 0.001), & to which it is much more vulnerable than XFG (its primary competitor in NA/Europe), is ~entirely escaped by A475V, judging by LF.7 vs LF.7.9 readings. Both images from Yunlong Cao studies.

131

New paper from phylogenetics/evolution arch-guru Richard Neher (@neher.io) and Luca Sesta! Tantalizing title ("Epistasis and the changing fitness landscapes of SARS-CoV-2") and abstract. Paywalled, unfortunately. academic.oup.com/genetics/adv...

1218

Here's a quick visual summary of the amino acid mutations on the branches leading from BA.2.86 to PJ.2.1.

1186

The situation is less clear for PJ.2.1. Compared to other circulating variants, it is more susceptible to class 3 and class 5 antibodies. These tend to be Omicron-specific, but I cannot find the exact epitopes of the mAbs highlighted in magenta below, so I can't make any precise predictions.

2202

What mutations should we expect PJ.2.1 to add? In late 2023, BA.2.86 had a clear Achilles Heel for type 1 Abs. @solidevidence.bsky.social famously predicted where it would add mutations (S:455-456). It did so promptly (L455S) becoming JN.1, which drove a large winter wave & still reigns worldwide.

1214

But as Yunlong Cao's team has shown, PJ.2.1 is similar to BA.2.75 or BA.2.86 in that it burst onto the scene with very high ACE2 affinity. (Lower value means stringer binding in graph below.) www.biorxiv.org/content/10.6...

1181

As @siamosolocani.bsky.social points out, @solidevidence.bsky.social's wastewater variant dashboard makes it clear that PJ.2.1 is the fastest-growing lineage in the US. I've inserted arrows pointing to some of the PJ.2.1 mutations. The farther above the dotted diagonal line, the faster the growth.

38525

Can't seem to get this to work. Does it have to be a GenBank fasta? I tried a regular GISAID fasta & it didn't work (got the top error message). I used Nextclade to get an aligned fasta but got the bottom error message. Genbank is basically useless to me. Impossible to find individual seqs there.

130

It's part of a larger mutation pattern that occurs in chronic sequences that includes ORF1a:233, ORF1a:1272-1273, and ORF1a:1361-1372. I'm not sure if this pattern is related to the Mac1-MAC2/neutrophil-mediated inflammatory reactions, but it's worth thinking about.

1111

Mutations from ORF1a:1361-1372 are enormously overrepresented in chronic-infection sequences, but even within this region, ORF1a:S1372 mutations—including S1372A, S1372C, S1372P, & S1372V—stand out.

1101

Among mutations occurring ≥25 times in a chronic-infection seq database I curate (called EPCI), ORF1a:S1372A is the 9th most over-represented mutation (relative to high-quality circulating seqs—HQCS). It's #4 among muts occurring ≥50 times. (6 of the top 8 are part of a deep-lung/BAL mut pattern)

180

...that seems important for the Mac1-Mac2/HEBP2 interaction is sometimes mutated in SARS-CoV-2, and that virtually exclusively during chronic infections. That's ORF1a:S1372 or NSP3:S554, which was mutated in the saltation variant BA.2.87.1 to S1372A.

182

And since HEBP2 suppresses inflammatory innate immune reactions, the result is excessive inflammation, NETs, etc. Mostly governed by highly conserved AA residues, which they demonstrated by mutagenesis are essential for the interaction. Only one AA residue... www.sciencedirect.com/science/arti...

162

A recent study from @cellpress.bsky.social's Cell Reports convincingly traces neutrophil inflammation and NETs during SARS-CoV-2 infection to the Mac2-Mac3 domains (which they call SUD2core) of NSP3. Mac2/3 recruits HEPB2 along with an E3 ligase that ubiquitinates it, leading to its degradation...

2247

Partly this is a raptor thing in general, but the northern goshawk in particular has always looked severely pissed off in nearly every picture I've seen of one. (Several times I've thought I saw one IRL, but never with certainty.) This picture from www.facebook.com/photo/?fbid=...

171

Or possibly AACGAACTAAA? That is very similar to the SARS-2 TRS, which is AAACGAAC (followed by TTTAA instead of TAAAA). The match with spike is pretty bad, but maybe very little spike is expressed in this virus.

120

Let's hope the AI bubble bursts as quickly as possible. Everyone knows it's a massive bubble—including AI agents. It's just a question of how long the collective insanity can persist. Below: ChatGPT's answer to @deanbaker13.bsky.social's question about the AI bubble. cepr.net/publications...

2154

I include the extra ORF8b (analog of ORF9b in SARS-2 or the "I" protein in other BetaCoVs) Kozaks because in MERS (most similar virus to BRZ), it's known that multiple proteins are produced by internal Ms in ORF8b. As in other BetaCoVs, BRZ has a lot of Ms in ORF8b, all with quite good Kozak seqs.

1113

I have no idea if the negative-strand ORFs here are real. However, searching for similarly large negative-strand ORFs in all betaCoV sequences, the only other comparably large one is a 250-AA one in mouse hepatitis virus (MHV-A59). When you search SARS-2 for (-)-strand ORFs, none reach 100 AA.

181

The green parts I put together/interpreted my self; the white parts are basically from the supplementary file from the study supplementary data table 2. "Kozak score" is my own invention based on the ideal human Kozak seq, worst possible being 0 and best 1. The extra Kozaks are for in-frame Ms.

140