Luis Zaman(@LuisZaman) 's Twitter Profileg
Luis Zaman

@LuisZaman

Assistant Prof. at University of Michigan in Complex Systems and Ecology & Evolutionary Biology. Find me on the other one, same username!

ID:51947712

linkhttp://www.zeeelab.com calendar_today29-06-2009 03:50:11

7,0K Tweets

3,5K Followers

2,2K Following

Dr. Emily Dolson(@emilyldolson) 's Twitter Profile Photo

I'll save the full tweetorial for Alex Lalejini and/or Luis Zaman but this is a really fun paper! Definitely check it out if you're interested in either of the following:

1. Steering evolution
2. Cross-talk between evolutionary computation and evolutionary biology

account_circle
Luis Zaman(@LuisZaman) 's Twitter Profile Photo

There is *much* more in the paper, so please check it out if you're interested in learning more or implementing some of these approaches in your own work.

I'm hopeful that the cross-talk between directed evolution and evolutionary computation will continue!!

13/end

account_circle
Luis Zaman(@LuisZaman) 's Twitter Profile Photo

And even when the objectives targeted by directed evolution are aligned with individual fitness (i.e., natural selection is efficient), these algorithms still provide large benefits for the metapopulation and the phenotypic spread of evolved populations!

12/n

And even when the objectives targeted by directed evolution are aligned with individual fitness (i.e., natural selection is efficient), these algorithms still provide large benefits for the metapopulation and the phenotypic spread of evolved populations! 12/n
account_circle
Luis Zaman(@LuisZaman) 's Twitter Profile Photo

They also produce the most diverse sets of phenotypes within subpopulations, and the most 'spread' through the space of possible phenotypes. This could be extremely valuable when trying to evolve novel therapeutics, for example!!

11/n

account_circle
Luis Zaman(@LuisZaman) 's Twitter Profile Photo

But long story short, LEX and NDE outperform the more standard algorithms even when evaluations are done for at the subpopulation level. And they work quickly, outperforming others after only 10 rounds of selection!

10/n

But long story short, LEX and NDE outperform the more standard algorithms even when evaluations are done for at the subpopulation level. And they work quickly, outperforming others after only 10 rounds of selection! 10/n
account_circle
Luis Zaman(@LuisZaman) 's Twitter Profile Photo

Lexicase (LEX) and Non-Dominating Elite (NDE) selection use the multidimensionality of evaluation to their advantage in different ways. Some of the details are in the paper, and some I'd claim are still under active research!

9/n

account_circle
Luis Zaman(@LuisZaman) 's Twitter Profile Photo

Tournament selection is intuitive, you pick some individuals/populations at random, and the best one is propagated. Now, the more advanced algorithms don't aggregate the different facets of performance into a single fitness value, and make explicit use of this dimensionality

8/n

account_circle
Alita Burmeister(@AlitaBurmeister) 's Twitter Profile Photo

With the lab up and running, I'm looking for Ph.D. & Master's students for '23, and a postdoc or postbac to start anytime. Find more info on my website! 3/3

account_circle
Siobhán O'Brien(@Siderophile) 's Twitter Profile Photo

It's also the first definition of microbial evolution that appears on the nature portfolio website... hmm nature.com/subjects/bacte…

account_circle
William Harcombe(@HarcombeLab) 's Twitter Profile Photo

I can’t say enough great things about CABmicrobe! Amazing science and fun people. Super excited to get started on new collaborations.

Thank you María RG, Alejandra Rodriguez, Ana E. Escalante, Eria Rebollar and all the others that made this happen.

I can’t say enough great things about @CABmicrobe! Amazing science and fun people. Super excited to get started on new collaborations. Thank you @mrebolleda, @jandraRV, @anaelena2777, Eria Rebollar and all the others that made this happen.
account_circle