mSplicer - Improving the C. elegans genome annotation using machine learning
by Gunnar Rätsch, Sören Sonnenburg, Jagan Srinivasan, Hanh Witte, Klaus-Robert Müller, Ralf Sommer and Bernhard Schölkopf.
Published in PLoS Computational Biology, February, 2007. The paper is available for download here.
For modern biology, precise genome annotations are of prime importance as they allow the accurate definition of genic regions. We employ state of the art machine learning methods to assay and improve the accuracy of the genome annotation of the nematode Caenorhabditis elegans. The proposed machine learning system is trained to recognize exons and introns on the unspliced mRNA utilizing recent advances in support vector machines and label sequence learning. In 87% (coding and untranslated regions) and 95% (coding regions only) of all genes tested in several out-of-sample evaluations, our method correctly identified all exons and introns. Notably, only 37% and 50%, respectively, of the presently unconfirmed genes in the C. elegans genome annotation agree with our predictions, thus we hypothesize that a sizable fraction of those genes are not correctly annotated. A retrospective evaluation of the Wormbase WS120 annotation  of C. elegans reveals that splice form predictions on unconfirmed genes in WS120 are inaccurate in about 18% of the considered cases, while our predictions deviate from the truth only in 10 − 13%. We experimentally analyzed 20 controversial genes on which our system and the annotation disagree, confirming the superiority of our predictions. While our method correctly predicted 75% of those cases, the standard annotation was never completely correct. The accuracy of our system is further corroborated by a comparison with two other recently proposed systems that can be used for splice form prediction: SNAP and ExonHunter. We conclude that the genome annotation of C. elegans and other organisms can be greatly enhanced using modern machine learning technology.
Training the mSplicer involves solving a relatively large linear optimization problem, which we have implemented in MATLAB using the CPLEX optimization package. Additionally we have developed a standalone tool for predicting the splice form for C. elegans sequences implemented in PYTHON and C++ available under the GNU General Public License. It is based on python scripts that call methods implemented in C++ for predicting splice sites using Support Vector Machines  and Dynamic Programming for splice form prediction. These routines are part of the freely available Shogun toolbox for large scale kernel learning .
You may download the PYTHON and C++ based part of mSplicer here. Note that you need at least a python2.4 runtime environment to run the Linux executables. Given a python2.4, numpy (>=1.0) and Shogun toolbox (shogun-python-modular) installation running mSplicer via python on any platform is easy. Note that shogun will make use of multiple processors if available.
- mSplicer version 0.3
- mSplicer version 0.2
- msplicer data file 0.2 (36M) (required by all 3 packages below, extract such that the 'data' directory is a subdirectory of msplicer)
- PYTHON source package 0.2 (23K) (requires Shogun toolbox version 0.3.2 or later)
- Linux executable 0.2 (9M debian etch-i386 standalone, requires a gcc4.x compiled python2.4 runtime environment, i.e. libpython2.4.so.1, libc >= 2.3 )
- 64bit Linux executable 0.2 (13M debian sarge-amd64 standalone, requires a gcc3.x python2.4 runtime environment)
- mSplicer version 0.1
- PYTHON source package 0.1 (25M) (requires Shogun toolbox version 0.3.0 or later)
- Linux executable 0.1 (34M debian etch-i386 standalone, requires a gcc4.x compiled python2.4 runtime environment, i.e. libpython2.4.so.1, libc >= 2.3 )
- 64bit Linux executable 0.1 (39M debian sarge-amd64 standalone, requires a gcc3.x python2.4 runtime environment)
Following a statistical setup common in machine learning,3 we trained our system on 60% of the available cDNA sequences currently known for C. elegans (based on Wormbase , version WS120). The remaining 40% of the cDNA sequences were used to generate an independent set for out-of-sample testing. Additionally, we used available EST sequences (dbEST , as of 19/02/2004) to maximally extend the cDNA sequences at the 5’ and 3’ ends. For training, we did not use any EST sequences overlapping with the 40% of the cDNA sequences for out-of-sample prediction.
We provide the training, validation and test data sets that we have used for generating our results:
- Sequences for results based on WS120: elegans_WS120_mSplicer_sequences.tar.gz
- Sequences for results based on WS150: elegans_WS150_mSplicer_sequences.tar.gz
- Sequences for results based on WS160: elegans_WS160_mSplicer_sequences.tar.gz
Note that we omitted a few sequences when training with ORFs. Please contact Gunnar Rätsch for details.
We trained mSplicer based on information available with release WS160 of Wormbase and applied it to predict the splice form within the boundaries of annotated transcripts (WS160). The following GFF files are available for download:
- Normal predictions (can be viewed on http://www.wormbase.org; tracks mSplicer and mSplicer-ORF)
- Sensitive predictions
Please note that we have continued working on extending mSplicer and came up with a version that can predict whole gene structures. The new tool is called mGene and was developed for the nGASP competition.
|||Harris T, Chen N, Cunningham F, et al. (2004) Wormbase, a multi-species resource for nematode biology and genomics. Nucl Acids Res 32. D411-7.|
|||Cortes, C, Vapnik, VN. Support-vector networks. Machine Learning, 20(3):273--297, 1995.|
|||Sonnenburg, S, Rätsch, G, Schäfer, C, Schölkopf, B. Large Scale Multiple Kernel Learning. Journal of Machine Learning Research,7:1531-1565, July 2006, K.Bennett and E.P.-Hernandez Editors.|
|||Rätsch, G, Sonnenburg, S, Srinivasan, J, Witte, H, Müller, KR, Sommer, R, and Schölkopf, B (2007). Improving the C. elegans genome annotation using machine learning. PLoS Computational Biology 3(2):e20.|
|||Schwarz E, Antoshechkin I, Bastiani C, et al (2006) Wormbase, better software, richer content. Nucleic Acids Res 34:D475–8.|
|||Boguski M, Tolstoshev TLC (1993). dbEST–database for expressed sequence tags. Nat Genet 4,332–3.|