Our new paper in @currentbiology.bsky.social, led by @renatoamorais.bsky.social, examined the history and veracity of the alleged paradox and its patron Charles Darwin.

www.cell.com/current-biol...

@oclaripv.bsky.social
@paulinenarvaez.bsky.social
@utaustin.bsky.social
@psl-univ.bsky.social

This is a very informative read: Seven data-driven lessons from the 2025 elections
open.substack.com/pub/gelliott...

Omg I’m dying 😂😂😂…so…So Emily Austin, a right wing 'influencer' who works for Bibi and the GOP, launched 'Hot Girls for Cuomo'.

Guess who didn't bother to register the URL when she did that? What do you think Hot Girls for Cuomo dot com leads to now?... 😄 Enjoy.

[ HotGirlsforCuomo.com ]

Amen. I tend to default to “the internet was a bad idea,” but I think it’s more specifically “engagement algorithms are a terrible idea and should probably be outlawed entirely for the public good,” and I’m not joking.

I feel like PCAs might be misleading. Like, so many cases of “yeah, that point X closer to point Y but that doesn’t mean it’s not correlated to Z further away, X is just more similar to Y than to Z.”

To me this feels like a network/WGCNA sort of thing, if you have someone that can do it? You can identify clusters of correlated microbes based on abundance.

Mixomics might have a method for this, too? What be a good question for Kim Ann regardless.

Oh that’s good to know! Any idea of proportions of chimeras vs duplex for those?

No one at the time I was investigating knew much about mid-strand barcodes and they were few, so throwing out was the safest option. For me that’s the case. If they can be safely handled another way, seems useful.

Indeed does happen, but more often they are duplex reads that are not detected as such by Dorado. Then the sequences on either side of the mid strand adapter are reverse complement (example pic)

Summary: de facto replacing PoreChop with a new adapter/barcode trimmer seems like a good idea. PoreChop isn’t maintained anymore but people are still using it. Would be useful to have a current tool.

If splitting, if you wanted to be fancy, you could scan both ends for barcodes and, once split, repatriate the ends to the correct sample if they’re different, which would be cool, but in practice they’re so few reads that it’s likely not worth the extra compute to do. I just chuck them out.

By definition, my understanding is that reads with mid-strand adapters are chimeras generated by one read sticking to another as they go through the poor, generating a concatenated read. That could pull sequence from one sample into another and you’d want to just get rid of those.