[3dem] ModelAngelo: automated model building

[Sjors Scheres - MRC LMB]

 Dear EM-ers,

We are proud to present ModelAngelo: a deep-learning approach for
automated model building in cryo-EM maps. In our experience, models
built by #ModelAngelo approximate those built manually for maps
extending beyond 3.5A resolution. Users can provide a FASTA sequence
file containing all chains that are present in the map to aid building.
Alternatively, for maps with unknown sequences, it can output an HMM
profile to search a sequence database, e.g. using HHBlits, to identify
the proteins present.

Preprint (for a machine-learning audience):
https://urldefense.com/v3/__https://arxiv.org/pdf/2210.00006.pdf__;!!Mih3wA!Gm4B08x1_58daVS-_1VeTE8trgyKjvaugKKrNX0jkNM8__Z5A-UokIgiSlt5NMz-xqZnSnFJKVRF-5mSyNLMZrRtGEA$  .

Open-source code: https://urldefense.com/v3/__https://github.com/3dem/model-angelo__;!!Mih3wA!Gm4B08x1_58daVS-_1VeTE8trgyKjvaugKKrNX0jkNM8__Z5A-UokIgiSlt5NMz-xqZnSnFJKVRF-5mSyNLMDYLbez8$  

We are now preparing a paper for a structural biology audience too. If
you obtain exciting results with ModelAngelo, do let us know: we might
want to write about it! Kiarash (kjamali at mrc-lmb.cam.ac.uk) is also keen
on helping out with cases of identifying unknown proteins in the map.

Have fun!

Kiarash Jamali, Dari Kimanius & Sjors Scheres


-- 
Sjors Scheres
MRC Laboratory of Molecular Biology
Francis Crick Avenue, Cambridge Biomedical Campus
Cambridge CB2 0QH, U.K.
tel: +44 (0)1223 267061
https://urldefense.com/v3/__http://www2.mrc-lmb.cam.ac.uk/groups/scheres__;!!Mih3wA!Gm4B08x1_58daVS-_1VeTE8trgyKjvaugKKrNX0jkNM8__Z5A-UokIgiSlt5NMz-xqZnSnFJKVRF-5mSyNLMovpnjMo$