IRENE Conference

  • Published on
    05-Dec-2014

  • View
    747

  • Download
    0

Embed Size (px)

DESCRIPTION

Recording of the talk: http://proteinsandwavefunctions.blogspot.com/2012/05/my-talk-at-irene-meeting.html

Transcript

<ul><li> 1. Industrial Strength QM/MM: Computa8onal high throughput screening of enzyme ac8vity in enzyme mutants Jan H. Jensen, Mar$n Hediger, Luca De Vico, Kasper Primdal, Allan Svendsen, Werner Besenma=er Department of Chemistry University of Copenhagen Slides at: h=p://Fnyurl.com/bsqbojf MarFn R. Hediger, Luca De Vico, Allan Svendsen, Werner Besenma=er, Jan H. Jensen A ComputaFonal Methodology to Screen AcFviFes of Enzyme Variants PLoS ONE, submi=ed. h=p://arxiv.org/abs/1203.2950 </li> <li> 2. Slides at: h=p://Fnyurl.com/bsqbojf Industrial enzyme design High-through put screening of 100s of mutants IdenFes promising candidates for further study ComputaFonal predicFon: Homology modeling QSAR (QM or QM/MM too slow and lacks automaFon) IdenFes promising candidates for further study Further study: 20-50 mutants Goal Automated predicFon of barrier height for enzymaFc reacFon within 24 hr using &lt; 10 cores IdenFes promising candidates for further study </li> <li> 3. Methods PM6 implemented in Mopac2009 (MOZYME) Automated mutant builder (PYMOL) Barrier from adiabaFc mapping Applica$on Increase amidase acFvity in an estarase (CalB) MarFn R. Hediger, Luca De Vico, Allan Svendsen, Werner Besenma=er, Jan H. Jensen A ComputaFonal Methodology to Screen AcFviFes of Enzyme Variants PLoS ONE, submi=ed. </li> <li> 4. PM6 is good enough </li> <li> 5. PM6 and MOZYME MOZYME = PM6 computed with MOZYME PM6 = PM6//MOZYME MOZYMEReortho = MOZYME//MOZYME </li> <li> 6. PM6/MOZYME is fast enough 55 aa MOPAC2009 No parallelized </li> <li> 7. PM6/MOZYME is fast enough OpFmizaFon Single point </li> <li> 8. Future Direc$ons Whole protein COSMO solvaFon More automaFzaFon Be=er sampling Complete scan of single mutants Single -&gt; double -&gt; triple mutants PM6 in GAMESS Linear scaling PM6 PM6/PCM interface AlternaFves to adiabaFc mapping Beyond PM6: EFMO </li> <li> 9. Blurring the boundary between linear scaling QM, QM/MM and polarizable force elds The Eec@ve Fragment Molecular Orbital Method Jan H. Jensen, Casper Steinmann, Mikael Wistoi Ibsen, Kasper Thoie University of Copenhagen Dmitri Fedorov AIST, Japan Casper Steinmann, Dmitri G. Fedorov, and Jan H. Jensen The EecFve Fragment Molecular Orbital Method: A Merger of the Fragment Molecular Orbital and EecFve Fragment PotenFal Methods Journal of Physical Chemistry A 2010, 114, 8705-8712 Casper Steinmann, Dmitri G. Fedorov, and Jan H. Jensen The EecFve Fragment Molecular Orbital Method for Fragments Connected by Covalent Bonds PLoS ONE, submi=ed. h=p://arxiv.org/abs/1202.4935 11 </li> <li> 10. The Eec$ve Fragment Molecular Orbital (EFMO) method Using ideas from the EecFve Fragment PotenFal (EFP) and the Fragment Molecular Orbital (FMO) method 12 </li> <li> 11. The Eec$ve Fragment Molecular Orbital (EFMO) method (Using ideas from the EecFve Fragment PotenFal (EFP) method) Monomer SCF in the gas phase Extract mulFpoles and dipole polarizability 13 </li> <li> 12. The Eec$ve Fragment Molecular Orbital (EFMO) method (Using ideas from the EecFve Fragment PotenFal (EFP) method) Many-body polariza$on Computed classically using induced dipoles for enFre system 14 </li> <li> 13. The Eec$ve Fragment Molecular Orbital (EFMO) method (Using ideas from the EecFve Fragment PotenFal (EFP) method) Coulomb and Non-Coulomb eects dimer SCF in the gas phase 15 </li> <li> 14. The Eec$ve Fragment Molecular Orbital (EFMO) method (Using ideas from the EecFve Fragment PotenFal (EFP) method) Coulomb eects Computed using staFc mulFpoles 16 </li> <li> 15. MP2 (DFT doesnt scale well) + 0 17 </li> <li> 16. Covalent Fragmenta$on (ElectrostaFc screening crucial) 18 </li> <li> 17. Implemented in GAMESS With gradients Trp cage (20 residues) 2 residues/fragment EFMO FMO2 Error in energy -4.3 6.4 kcal/mol MP2/6-31G(d) gradient 314 409 minutes 20 cores (most Fme spent in MP2 dimers) 19 </li> <li> 18. QM/MM PCM Large parts of MM region oien frozen = Requires only monomer gas phase calculaFons for that region = Very fast 20 </li> <li> 19. To Do Flexible EFP/Polarizable Force Field covalent dimers (E ) NE EFMO = EI0 + 0 IJ EI0 EJ EIJ 0 POL I IJ ( ) N + EIJ + EIJ /CT + EIJ + Etot ES XR Disp POL IJ Important miscellanea EFMO GUI: FRAGIT (Mikael Ibsen) TS search algorithms (Kasper Thoie) 21 </li> <li> 20. Funding: EU (IRENE collab program) Thank You! Ques$ons? Slides at: h=p://Fnyurl.com/bsqbojf 22 </li> </ul>