GPU-based computing for cryo-electron microscopy

Dnr:

SNIC 2017/1-8

Type:

SNAC Medium

Principal Investigator:

Martin Hällberg

Affiliation:

Karolinska Institutet

Start Date:

2017-01-30

End Date:

2018-02-01

Primary Classification:

10601: Strukturbiologi

Secondary Classification:

10602: Biokemi och molekylärbiologi

Tertiary Classification:

10603: Biofysik

Webpage:

http://hallberglab.org/

Allocation

Abstract

Biological electron microscopy is currently going through a revolution in terms of usability and the attained results in terms of the resolution (~quality) of the resulting reconstruction. The current state-of-the-art programs RELION 2.0 and cryoSPARC are very GPU intensive and we have a large need for access to GPU equipped computing nodes to process data and perform reconstructions of data collected primarily at SciLife but also other resources available through EU-wide access programs. With electron microscopy programs that we will run on the cluster we will analyse biological supermolecules involved in mitochondrial RNA processing, nuclear pore transport relevant for cancer progression (KI) and membrane protein complexes involved in cellulose synthesis (KTH).