Department information

 

Kernel Machines, Pattern Analysis and Computational Biology

Mission

The group develops machine learning methods, models and tools for computational sciences, in particular computational biology. The methodological backbone of the group is kernel methods and regularized learning. The group particularly focusses in learning with multiple and structured targets, multiple views and ensembles. Applications of interest in computational biology include protein function and interaction prediction as well as molecular classification and identification.

Core competence

• Optimization algorithms (convex and combinatorial)
• Machine learning, kernel methods
• Kernel methods for structured (output) prediction
• Metabolomics and metabolic network analysis
• Function prediction

People

• Juho Rousu - Principal Investigator, Group Leader
• Jana Kludas - Postdoctoral researcher
• Esa Pitkänen - Postdoctoral researcher
• Markus Heinonen - PhD student
• Hongyu Su - PhD student
• Huibin Shen - MSc student
• Yvonne Herrmann - MSc student

News

• We are hiring! If you are looking for a Post-doc/PhD student position in kernel methods and computational biology, please contact the group leader.
• From beginning of 2012, the group will mainly based at Aalto University, Department of Information and Computer Science.

Activities

Current

• BIOLEDGE - BIO knowLEDGE Extractor and Modeller for Protein Production. EU FP7 STREP (2011-2016).
• Algorithmic Data Analysis Academy of Finland National Center of Excellence.
• GEOBIOINFO - Deep biosphere bioinformatics. Part of KYT2014 programme funded by Ministry of Employment and the Economy

Past

• Workshop on Mass Spectrometry Informatics in Systems Biology, Helsinki, October 28-29,2010
• UR-ENZYMES - Modelling functional shifts in enzyme evolution. Academy of Finland MASI programme.
• Experimental and computational analysis of physiological regulation at transcriptome, proteome and metabolome level (SYSFYS), Academy of Finland SYSBIO programme.
• Integrated Computational Methods for Genomic, Proteomic and Metabolic Modelling (ICOMIC), Academy of Finland MADAME programme.

Software

Publications

• Hongyu Su, Juho Rousu: Multi-Task Drug Bioactivity Classification with Graph Labeling Ensembles. Pattern Recognition in Bioinformatics, 2011, to appear
• Esa Pitkänen, Mikko Arvas, Juho Rousu: Minimum mutation algorithm for gapless metabolic network evolution. Bioinformatics Models, Methods and Algorithms, 2011.
• Hongyu Su, Markus Heinonen, Juho Rousu.Multilabel Classification of Drug-like Molecules via Max-Margin Conditional Random Fields. In Probabilistic Graphical Models, 2010. [dataset]
• Hongyu Su, Markus Heinonen, Juho Rousu. Structured Output Prediction of Anti-Cancer Drug Activity. Pattern Recognition in Bioinformatics 2010. Lecture notes in bioinformatics, 2010, to appear. Preliminary version
• Katja Astikainen, Liisa Holm, Esa Pitkanen, Sandor Szedmak and Juho Rousu. Structured Output Prediction of Novel Enzyme Function with Reaction Kernels. BIOSTEC 2010 Revised Selected Papers. Springer Communications in Computer and Information Science, to appear. Preliminary version
• Saso Dzeroski, Pierre Geurts, Juho Rousu. Proceedings of the Third International Workshop on Machine Learning in Systems Biology: Revised Selected Papers. Journal of Machine Learning Research Workshop and Conference Proceedings 8:1-2, 2010
• Markus Heinonen, Sampsa Lappalainen, Taneli Mielikäinen, Juho Rousu. Computing Atom Mappings for Biochemical Reactions without Subgraph Isomorphism. Journal of Computational Biology, 2010, to appear. Preliminary version
• Esa Pitkänen, Juho Rousu, Esko Ukkonen: Computational methods for metabolic reconstruction. Current Opinion in Biotechnology, 21(1):70-77, 2010.
• Katja Astikainen, Liisa Holm, Esa Pitkanen, Sandor Szedmak, Juho Rousu. Reaction kernels: structured output prediction approaches for novel enzyme function. 1st International Conference on Bioinformatics, Valencia, Spain, January 2010.
• Esa Pitkänen, Paula Jouhten, Juho Rousu: Inferring branching pathways in genome-scale metabolic networks. BMC Systems Biology, 3:103, 2009.
• Paula Jouhten, Esa Pitkänen, Tiina Pakula, Markku Saloheimo, Merja Penttilä, Hannu Maaheimo: 13C-metabolic flux ratio and novel carbon path analyses confirmed that Trichoderma reesei uses primarily the respirative pathway also on the preferred carbon source glucose. BMC Systems Biology, 3:104, 2009
• Sandor Szedmak, Craig Saunders, Yizhao Ni and Juho Rousu. Max-margin structured output learning in L1 norm space, PASCAL Technical reports, 2009
• Saso Dzeroski, Pierre Geurts, Juho Rousu (Eds.). Machine Learning in Systems Biology. Proc. 3rd International Workshop, Ljubljana, Slovenia, September 5-6, 2009. Report B-2009-1, Department of Computer Science, University of Helsinki
• Juho Rousu. Bioinformatics: Technologies and Challenges. Proceedings of AMICT 2008, Advances in Methods of Information and Communication Technology. Petrozavodsk State University, 2009, 50--57
• Esa Pitkänen, Arto Åkerlund, Ari Rantanen, Paula Jouhten, Esko Ukkonen: ReMatch: a web-based tool to construct, store and share stoichiometric metabolic models with carbon maps for metabolic flux analysis. Journal of Integrative Bioinformatics, 5(2):102, 2008.
• Esa Pitkänen, Ari Rantanen, Juho Rousu, Esko Ukkonen A computational method for reconstructing gapless metabolic networks. 2nd International Conference on Bioinformatics Research and Development (BIRD'08), Communications in Computer and Information Science, Vol. 13, Springer, 2008.
• Katja Astikainen, Liisa Holm, Esa Pitkänen, Sandor Szedmak, Juho Rousu. Towards Structured Output Prediction of Enzyme Function . BMC Proceedings 2, Suppl 4 (2008):S2
• Markus Heinonen, Ari Rantanen, Taneli Mielikäinen, Juha Kokkonen, Jari Kiuru, Raimo A. Ketola, Juho Rousu: FiD: a software for ab initio structural identification of product ions from tandem mass spectrometric data. Rapid Communications in Mass Spectrometry 22, 19 (2008), 3043 - 3052
• Ari Rantanen, Juho Rousu, Paula Jouhten, Nicola Zamboni, Hannu Maaheimo, Esko Ukkonen. An analytic and systematic framework for estimating metabolic flux ratios from 13C tracer experiments. BMC Bioinformatics 2008, 9:266.
• J. Rissanen, P. Grunwald, J. Heikkonen, P. Myllymäki, T Roos, Juho Rousu (eds.). Information Theoretic Methods for Bioinformatics. EURASIP Journal on Bioinformatics and Systems Biology, vol. 2007, (2007)
• Samuel Kaski, Esko Ukkonen, Juho Rousu (eds.) Probabilistic modeling and machine learning in structural and systems biology . PMSB 2006 special issue. BMC Bioinformatics 2007, 8 (Suppl 2), S1
• Huizhen Yu, Juho Rousu. An Efficient Method for Large Margin Parameter Optimization in Structured Prediction Problems , Technical report C-2007-87, Dept. Computer Science, Univ. of Helsinki, 2007.
• Juho Rousu, Craig Saunders, Sandor Szedmak, John Shawe-Taylor. Efficient algorithms for max-margin structured classification. In G. Bakir, T. Hofmann, B. Schölkopf, A. Smola, B. Taskar, S.V.N Vishwanathan (eds.): Predicting Structured Data, MIT Press, 2007
• Juho Rousu, Craig Saunders, Sandor Szedmak and John Shawe-Taylor. Kernel-based Learning of Hierarchical Multilabel Classification Models . Journal of Machine Learning Research 7 (2006), pp. 1601 - 1626. MATLAB implementation of the learning algorithm is available here.
• Markus Heinonen, Ari Rantanen, Taneli Mielikäinen, Esa Pitkänen, Juha Kokkonen, Juho Rousu: Ab Initio Prediction of Molecular Fragments from Tandem Mass Spectroscopy Data. German Conference on Bioinformatics 2006 (GCB 2006). Lecture Notes in Informatics Vol. P-83 (2006), pp. 40-53.
• Pekko Parikka, Esa Pitkänen, Ari Rantanen, Arto Åkerlund, Esko Ukkonen: Pathway Assistant: a web portal for metabolic modelling. Network Tools and Applications in Biology (NETTAB'06), pp. 90 - 96, 2006.
• Ari Rantanen, Taneli Mielikäinen, Juho Rousu, Hannu Maaheimo, Esko Ukkonen: Planning optimal measurements of isotopomer distributions for estimation of metabolic fluxes (extended version of GCB2005 manuscript). Bioinformatics, 22(10), pp. 1198 - 1206, 2006.
• Ari Rantanen, Taneli Mielikäinen, Juho Rousu and Esko Ukkonen. Planning isotopomer measurements for estimation of metabolic fluxes. German Conference on Bioinformatics 2005, Lecture Notes in Informatics vol. P-71, pp. 177 - 191, 2005
• Esa Pitkänen, Ari Rantanen, Juho Rousu and Esko Ukkonen. Finding Feasible Pathways in Metabolic Networks. Advances in Informatics: 10th Panhellenic Conference on Informatics (PCI 2005), Lecture Notes in Computer Science 3746, pp. 123 - 133, 2005.
• Ari Rantanen, Juho Rousu, Esa Pitkänen, Hannu Maaheimo and Esko Ukkonen. Flow analysis of metabolite fragments for flux estimation. Computational Methods in Systems Biology, Edinburgh, Scotland, pp. 242 - 255, 2005.
  Supplementary material: Metabolic network, dominator trees.
• Juho Rousu, Ari Rantanen, Raimo A. Ketola and Juha T. Kokkonen. Isotopomer distribution computation from tandem mass spectrometric data with overlapping fragment spectra. Spectroscopy 19(1), pp. 53 - 67, 2005.
• Juho Rousu, John Shawe-Taylor. Efficient Computation of Gapped Substring Kernels on Large Alphabets. Journal of Machine Learning Research 6 (2005), 1323-1344. MATLAB Implementation of the sparse dynamic programming algorithm

---

Previous update: 16.12.2011 - Juho Rousu