MStream: proof of concept of an analytic cloud platform for near-real-time diagnostics using mass spectrometry data / Roman Zoun, Kay Schallert, David Broneske, Sören Falkenberg, Robert Heyer, Sabine Wehnert, Sven Brehmer, Dirk Benndorf and Gunter Saake (Arbeitsgruppe DBSE)
Anzeigen / Download1.08 MB
Discovery
1677331879
URN
urn:nbn:de:gbv:3:2-112088
DOI
ISBN
ISSN
Beiträger
Körperschaft
Erschienen
Magdeburg : Fakultät für Informatik, Otto-von-Guericke-Universität Magdeburg, 2019
Umfang
1 Online-Ressource (11 ungezählte Seiten, 1,08 MB) : Illustrationen, Diagramme
Ausgabevermerk
Sprache
eng
Anmerkungen
Inhaltliche Zusammenfassung
A mass spectrometer is a device to extract biomarkers of biological environments. Using these biomarkers, it is possible to diagnose thousands of diseases with only one mass spectrometer. Unfortunately, the mass spectrometry pipeline is sequential, including hours of waiting time between the workflow steps. Additionally, the data analysis is complex and needs qualified employees and a stable infrastructure, which involves very high costs and effort. Hence, only few hospitals use a mass spectrometer for diagnostics with success. In our work, we present a proof of concept of an analytical platform for real-time analysis of mass spectrometry experiments. In collaboration with Bruker Daltonik GmbH, we implemented MStream, a cloud-based platform on the SMACK stack (Spark, Mesos, Akka, Cassandra, Kafka) for scalable, streamlined protein identification. Our evaluation shows superior performance in comparison to the state-of-the-art X!Tandem software package. Additionally, we minimize the effort of the hospital by allowing the full analysis pipeline to be outsourced to our cloud platform.
Schriftenreihe
Technical report ; 002-2019 ppn:570164265