News

We are looking for PhD students, postdocs, Master,
Bachelor and Hiwi students!

November, 2013


Zwei Masterstudenten fuer spannende Projekte gesucht!

November, 2013

Aktuell suchen wir zwei Masterstudenten (evt. Bachelor) fuer zwei spannende Projekte im Bereich der Proteinstrukturvorhersage. Es geht um die Vorhersage von Kontakten zwischen Aminosäuren in Proteinen aufgrund ihrer korrelierten Mutationen. Diese Kontakte können zur de-novo Proteinstrukturvorhersage (also ohne bekannte, homologe Strukturen) oder in Verbindung mit Homologiemodellierung genutzt werden. Seit einem methodischen Durchbruch vor 2 Jahren herrscht weltweit großes Interesse an diesen Methoden. Unsere Gruppe forscht seit 1 1/2 Jahren auf dem Gebiet und hat vielversprechende Ideen, die wir gern schnell umsetzen möchten, um unsere eigene Implementierung voranzutreiben.

Im ersten Projekt geht es um die statistische Beschreibung, welche Paare von Aminosäuren äquivalent fuer Kontakbildung und Stabilitaet innerhalb der Protein sind. Eine Beschreibung mit Hilfe statistischer Modellierung und Machine Learning könnte die Vorhersagen deutlich verbessern. Im zweiten Projekt verfolgen wir den Ansatz, Muster in der Kontaktmatrix zu erkennen, die fuer den Kontakt bestimmter Sekundärstrukturelemente charakteristisch sind. Die Arbeiten sind ideal fuer Leute mit Spass an quantitativem Denken und Statistik. Gute Programmierkenntnisse wären vorteilhaft. Die Arbeiten umfassen die gesamte Entwicklung der jeweiligen Methoden, von der Generierung von Trainings- und Testdaten ueber die statistische Modellierung (theoretisch und praktische Umsetzung) und Benchmarking der Methode und die iterative Verbesserung. Falls ihr Interesse habt, meldet euch bei bei Johannes Soeding (soeding@genzentrum.lmu.de). Ihr koennt euch natuerlich auch gern an Gruppenmitglieder mit Fragen zur Gruppe wenden (Liste unter "People").


Lab move to Goettingen sealed!

November, 2013

Our lab is going to move to the Max-Planck Institute for Biophysical Chemistry in Goettingen in July 2014. Official start is already January 1, 2014, although our office space will be ready only in July. The MPI-BPC is the largest institute of the Max-Planck society and one of the most successfull ones (sporting, among other prize winners, two nobelists). It will offer excellent working conditions and exciting new opportunities for collaborations with world-leading groups in cell and developmental biology, biochemistry and biophysics.


Jessica wins Humboldt fellowship

November, 2013

Jessica Andreani has earned a Humboldt fellowship to support her work on the de-novo prediction of residue-residue contacts in proteins from correlated mutations.


We welcome our new postdoc Jessica Andreani to our group!

September, 2013

Jessica (http://jessica.andreani.info/) comes from the École de Mines in Paris and will work on the de-novo prediction of residue-residue contacts in proteins from correlated mutations in a team with Stefan Seemayer and Susann Vorberg. She will develop probablistic methods to predict contacting secondary structure elements in proteins.


Manuscript on de-novo motif finding accepted by Genome Research

September, 2012

Link zur Pressemitteilung             Link to press release

Our method XXmotif uses a new approach for finding enriched motifs in a set of sequences. As the classic tool MEME, it represents motifs by positional weight matrices (PWMs), but instead of maximizing a likelihood, it directly maximizes the enrichment P-values of candidate PWMs. The sensitivity for detecting weak motifs and the accuracy of reported PWMs were found to be superior to state-of-the-art tools. Genome Research, in press..


Our third PhD Student, Holger Hartmann graduates.

Congratulations!

May, 2012

Holger Hartmann, the Söding group's third PhD student, graduated on May 8 with highest distinction at the faculty of Chemistry and Pharmacy of the LMU Munich. The title of his dissertation is "Regulatory motif discovery using PWMs and the architecture of eukaryotic core promoters". Since April 1, Holger works for the German Chemistry giant BASF in Ludwigshafen, Germany, as one of their first computational biologists. Pioneering and exciting times ahead for him. Good luck, Holger.


New open-source release 2.0.0 of HH-suite available

Januray, 2012

The HH-suite 2.0.0 is an open-source software package for sensitive sequence searching based on the pairwise alignment of hidden Markov models (HMMs). It contains HHsearch and HHblits among other programs and utilities. Download HH-suite here.


HHblits manuscript accepted by Nature Methods

December, 2011

Link zur Pressemitteilung             Link to press release

The manuscript describing our iterative protein sequence search method HHblits (HMM-HMM-based lightning-fast iterative sequence search) has been accepted by Nature Methods. HHblits is the fastest and most sensitive all purpose sequence search method available. It will improve many downstream analysis and prediction methods that, so far, have relied on PSI-BLAST for building multiple sequence alignments.


Michael Remmert graduates. Congratulations!

November, 2011

Michael Remmert, developer of HHblits, the most sensitive protein sequence search method available, has graduated and left our group for new shores, which are fortunately not very distant. He now works with GeneData GmbH just next to our LMU Grosshadern/Martinsried campus.


Michael Remmert recieves ROMIUS Förderpreis

December, 2010

The ROMIUS Stiftung awards the Förderpreis für Wissenschaft und Forschung to Michael Remmert for outstanding scientific work in the field of the evolution of outer membrane proteins. "His research explains the evolutionary origin of outer membrane proteins, which lays the basis for the identification and classification of this important bacterial protein class." (from the prize statement).   


HHpred among top servers in CASP9

December, 2010

Official results of the community-wide protein structure prediction experiment CASP (Critical Assessment of Techniques for Protein Structure Prediction) have now been released. Our HHpred server is ranked 1st in template-based modeling among 81 participating servers (CASP9 TBM). With a median response time of 4-5 min, HHpred is also one of the fastest server in CASP9. Important improvements with respect to the earlier version of CASP8 include building query and database multiple alignments with our new tool HHblits instead of PSI-BLAST and developing probabilistic atom-atom distance restraints that depend on the local alignment quality. 


Our first Phd student graduates with summa cum laude

December, 2010

Andreas Biegert, the Söding group's first PhD student, graduated on December 14 with highest distinction at the faculty of Chemistry and Pharmacy of the LMU Munich. The title of his dissertation is "Context-specific methods for sequence homology searching and alignment". 


Maria Hauser wins coveted Telekom stipend

December, 2010

Maria is one of this year's 12 PhD students to win a high-paying telekom stipend for her PhD project. She will develop methods to dissect the entire sequence space into functional, structural, and evloutionary domains, which promises to greatly facilitate the functional characterization of proteins and to shed light on the evolution of the protein universe. Congratulations! 

 


Laborjournal: HHsearch among top 10 publications in
evolutionary biology

July, 2010

Laborjournal: HHsearch among top 10 publications in evolutionary biology July, 2010
The 2005 Bioinformatics article describing our HMM-HMM comparison method HHsearch is ranked 9th in numbers of citations in the field of evolutionary biology by Laborjournal in its June issue. Johannes Soeding is among most highly cited researcher in this field in Germany, Austria, and Switzerland. See Laborjournal ranking


HHsearch and HHpred cited more than 500 times

March, 2010

HHsearch, our method for remote homology detection (Soeding 2005), now exceeds 300 citations and HHpred, the web server for structure prediction (Soeding et. al, 2005) is cited over 200 times in Google scholar


"Evolution of outer membrane β-barrels from an
ancestral ββ hairpin" 
published in Molecular Biology and Evolution

January, 2010

In this paper we show that the large and diverse group of outer membrane beta barrels, the major class of proteins in bacterial and mitochondrial outer membranes, is descended from a single ββ hairpin structure. This two-stranded hairpin module was amplified early in protein evolution to give rise to the 8 to 24-stranded outer membrane beta barrels. Using a combined sequence-structure analysis, we could for the first time exclude sequence convergence as explanation for the observed very weak sequence similarities. 


The Galaxy of folds  appears on the cover page of the
January 2010 issue of Protein Science

January, 2010

Despite their enormous sequence diversity, proteins are built of recurrent domains belonging to only a few thousand fold types. Did these folds originate independently multiple times, converging on similar structural solutions, or did they evolve divergently from ancestral prototypes? In this study, we clustered domains of known structure by their sequence similarity, a property that reflects common descent. While some highly populated folds indeed appear to have evolved convergently, most domains possessing the same fold arose from an ancestral prototype. Occasionally, domains belonging to different folds cluster together, connected by recurrent fragments which may be descendants of an ancestral pool of peptide modules from which the first folded proteins arose. 


HHomp Webserver

May 8, 2009

Our manuscript HHomp - prediction and classification of outer membrane proteins has been published in NAR. This paper describes a new method and webserver for the prediction and classification of outer membrane proteins by HMM-HMM comparison. 


Context-specific amino acid similarities

February 20, 2009

In our manuscript Sequence context-specific profiles for homology searching appearing in PNAS, we describe a method CS-BLAST that is able to boost the sensitivity of BLAST sequence searches by more than a factor of two without loss of speed or other disadvantages. Sequence searches are performed hundreds of thousands of times per day by biologists in order to predict the function and structure of proteins. The paradigm of sequence context specificity developed here is broadly applicable in biological sequence comparison. A PDF of the paper can be downloaded from here.

This week in PNAS
Press release: Relationships in Rank and File
Pressemitteilung: Mit der Verwandtschaft in Reih und Glied
Evaluated as exceptional in Faculty 1000