Sven Rahmann

Snakemake—a scalable bioinformatics workflow engine

SUMMARY Snakemake is a workflow engine that provides a readable Python-based workflow definition language and a powerful execution environment that scales from single-core workstations to compute clusters without modifying the workflow. It is the first system to support the use of automatically inferred multiple named wildcards (or variables) in input and output filenames. AVAILABILITY http://snakemake.googlecode.com. CONTACT johannes.koester@uni-due.de.

HMM Logos for visualization of protein families

BackgroundProfile Hidden Markov Models (pHMMs) are a widely used tool for protein family research. Up to now, however, there exists no method to visualize all of their central aspects graphically in an intuitively understandable way.ResultsWe present a visualization method that incorporates both emission and transition probabilities of the pHMM, thus extending sequence logos introduced by Schneider and Stephens. For each emitting state of the pHMM, we display a stack of letters. The stack height is determined by the deviation of the positions letter emission frequencies from the background frequencies. The stack width visualizes both the probability of reaching the state (the hitting probability) and the expected number of letters the state emits during a pass through the model (the states expected contribution).A web interface offering online creation of HMM Logos and the corresponding source code can be found at the Logos web server of the Max Planck Institute for Molecular Genetics http://logos.molgen.mpg.de.ConclusionsWe demonstrate that HMM Logos can be a useful tool for the biologist: We use them to highlight differences between two homologous subfamilies of GTPases, Rab and Ras, and we show that they are able to indicate structural elements of Ras.

Exome sequencing identifies recurrent somatic mutations in EIF1AX and SF3B1 in uveal melanoma with disomy 3

Gene expression profiles and chromosome 3 copy number divide uveal melanomas into two distinct classes correlating with prognosis. Using exome sequencing, we identified recurrent somatic mutations in EIF1AX and SF3B1, specifically occurring in uveal melanomas with disomy 3, which rarely metastasize. Targeted resequencing showed that 24 of 31 tumors with disomy 3 (77%) had mutations in either EIF1AX (15; 48%) or SF3B1 (9; 29%). Mutations were infrequent (2/35; 5.7%) in uveal melanomas with monosomy 3, which are associated with poor prognosis. Resequencing of 13 uveal melanomas with partial monosomy 3 identified 8 tumors with a mutation in either SF3B1 (7; 54%) or EIF1AX (1; 8%). All EIF1AX mutations caused in-frame changes affecting the N terminus of the protein, whereas 17 of 19 SF3B1 mutations encoded an alteration of Arg625. Resequencing of ten uveal melanomas with disomy 3 that developed metastases identified SF3B1 mutations in three tumors, none of which targeted Arg625.

Deep sequencing reveals differential expression of microRNAs in favorable versus unfavorable neuroblastoma

Small non-coding RNAs, in particular microRNAs(miRNAs), regulate fine-tuning of gene expression and can act as oncogenes or tumor suppressor genes. Differential miRNA expression has been reported to be of functional relevance for tumor biology. Using next-generation sequencing, the unbiased and absolute quantification of the small RNA transcriptome is now feasible. Neuroblastoma(NB) is an embryonal tumor with highly variable clinical course. We analyzed the small RNA transcriptomes of five favorable and five unfavorable NBs using SOLiD next-generation sequencing, generating a total of >188 000 000 reads. MiRNA expression profiles obtained by deep sequencing correlated well with real-time PCR data. Cluster analysis differentiated between favorable and unfavorable NBs, and the miRNA transcriptomes of these two groups were significantly different. Oncogenic miRNAs of the miR17-92 cluster and the miR-181 family were overexpressed in unfavorable NBs. In contrast, the putative tumor suppressive microRNAs, miR-542-5p and miR-628, were expressed in favorable NBs and virtually absent in unfavorable NBs. In-depth sequence analysis revealed extensive post-transcriptional miRNA editing. Of 13 identified novel miRNAs, three were further analyzed, and expression could be confirmed in a cohort of 70 NBs.

On the power of profiles for transcription factor binding site detection

Transcription factor binding site (TFBS) detection plays an important role in computational biology, with applications in gene finding and gene regulation. The sites are often modeled by gapless profiles, also known as position-weight matrices. Past research has focused on the significance of profile scores (the ability to avoid false positives), but this alone is not enough: The profile must also possess the power to detect the true positive signals. Several completed genomes are now available, and the search for TFBSs is moving to a large scale; so discriminating signal from noise becomes even more challenging.Since TFBS profiles are usually estimated from only a few experimentally confirmed instances, careful regularization is an important issue. We present a novel method that is well suited for this situation.We further develop measures that help in judging profile quality, based on both sensitivity and selectivity of a profile. It is shown that these quality measures can be efficiently computed, and we propose statistically well-founded methods to choose score thresholds.Our findings are applied to the TRANSFAC database of transcription factor binding sites. The results are disturbing: If we insist on a significance level of 5% in sequences of length 500, only 19% of the profiles detect a true signal instance with 95% success probability under varying background sequence compositions.

Mutational dynamics between primary and relapse neuroblastomas

Neuroblastoma is a malignancy of the developing sympathetic nervous system that is often lethal when relapse occurs. We here used whole-exome sequencing, mRNA expression profiling, array CGH and DNA methylation analysis to characterize 16 paired samples at diagnosis and relapse from individuals with neuroblastoma. The mutational burden significantly increased in relapsing tumors, accompanied by altered mutational signatures and reduced subclonal heterogeneity. Global allele frequencies at relapse indicated clonal mutation selection during disease progression. Promoter methylation patterns were consistent over disease course and were patient specific. Recurrent alterations at relapse included mutations in the putative CHD5 neuroblastoma tumor suppressor, chromosome 9p losses, DOCK8 mutations, inactivating mutations in PTPN14 and a relapse-specific activity pattern for the PTPN14 target YAP. Recurrent new mutations in HRAS, KRAS and genes mediating cell-cell interaction in 13 of 16 relapse tumors indicate disturbances in signaling pathways mediating mesenchymal transition. Our data shed light on genetic alteration frequency, identity and evolution in neuroblastoma.

Partitioning biological data with transitivity clustering.

1. Huisken, J., Swoger, J., Del Bene, F., Wittbrodt, J. & Stelzer, E.H. Science 305, 1007–1009 (2004). 2. Huisken, J. & Stainier, D.Y. Development 136, 1963–1975 (2009). 3. Lindeberg, T. J. Appl. Stat. 21, 224–270 (1994). 4. Fischler, M.A. & Bolles, R.C. Commun. ACM 24, 381–395 (1981). 5. Preibisch, S., Saalfeld, S. & Tomancak, P. Bioinformatics 25, 1463–1465 (2009). 6. Preibisch, S., Rohlfing, T., Hasak M.P. & Tomancak P. SPIE Medical Imaging 2008 (eds., Reinhardt, J.M. & Pluim, J.P.W.) 6914, 69140E-69140E-8 (2008). 7. Swoger, J. et al. Opt. Express 15, 8029–8042 (2007). (Supplementary Table 1). The average bead displacement and the ratio between correspondence candidates and true correspondences is a quantitative measure of the reconstruction success, which is crucial for automatic validation of registration results in long time-lapse recordings. The beads can be removed optically or computationally from the sample (Supplementary Methods). We applied the bead-based registration framework to SPIM recordings of early Drosophila melanogaster embryos, which are very challenging samples for multiview reconstruction owing to the scattering of the yolk that severely limits the overlap between views. We imaged Drosophila embryos expressing ubiquitous HisYFP from five and seven views in an extended time-lapse recording covering early embryonic development. We registered each time point separately and then registered all time points to each other compensating for minor drift during image acquisition (Supplementary Methods). We combined content-based fusion with nonlinear blending5 to compensate for brightness differences at boundaries between views (Supplementary Fig. 4). The reconstructed multiview acquisition of the specimen showed, in contrast to the single view, comparable lateral and axial resolution (Fig. 1e,f). We never imaged the anterior and posterior poles of the embryo with full lateral resolution in this acquisition, and yet the cells were clearly distinguishable, demonstrating the precision of the multiview reconstruction (Fig. 1g–i). In the middle of the specimen, the resolution was lower because only some views contributed high-content information whereas other views were blocked by the yolk (Fig. 1h). The reconstructed time-lapse recording provided an unprecedented four-dimensional view of Drosophila embryogenesis (Supplementary Videos 4 and 5). The bead-based registration framework is sample-independent (Supplementary Data and Supplementary Fig. 5) and enables fully unguided registration without prior knowledge of the arrangement of the views (Supplementary Video 4). The software outperforms intensity-based registration approaches6,7 in terms of precision and speed, enabling accurate registration of large, multiview acquisitions in minutes (Supplementary Data, Supplementary Fig. 6 and Supplementary Table 1). The run time of the bead-based registration framework is comparable to the time it takes to acquire the multiview data, and thus, to our knowledge, it is currently the only solution allowing robust, real-time registration of time-lapse SPIM recordings. Moreover, the bead-based registration framework is applicable to other optical sectioning microscopy techniques (Supplementary Fig. 7 and Supplementary Video 6), considerably expanding the possible applications in biology. We provide our bead-based registration algorithm to the bioimaging community as an open-source plugin for Fiji (Supplementary Fig. 8 and Supplementary Software; http://pacific.mpi-cbg.de/wiki/index. php/SPIM_Registration).

A comprehensive molecular study on Coffin–Siris and Nicolaides–Baraitser syndromes identifies a broad molecular and clinical spectrum converging on altered chromatin remodeling

Chromatin remodeling complexes are known to modify chemical marks on histones or to induce conformational changes in the chromatin in order to regulate transcription. De novo dominant mutations in different members of the SWI/SNF chromatin remodeling complex have recently been described in individuals with Coffin-Siris (CSS) and Nicolaides-Baraitser (NCBRS) syndromes. Using a combination of whole-exome sequencing, NGS-based sequencing of 23 SWI/SNF complex genes, and molecular karyotyping in 46 previously undescribed individuals with CSS and NCBRS, we identified a de novo 1-bp deletion (c.677delG, p.Gly226Glufs*53) and a de novo missense mutation (c.914G>T, p.Cys305Phe) in PHF6 in two individuals diagnosed with CSS. PHF6 interacts with the nucleosome remodeling and deacetylation (NuRD) complex implicating dysfunction of a second chromatin remodeling complex in the pathogenesis of CSS-like phenotypes. Altogether, we identified mutations in 60% of the studied individuals (28/46), located in the genes ARID1A, ARID1B, SMARCB1, SMARCE1, SMARCA2, and PHF6. We show that mutations in ARID1B are the main cause of CSS, accounting for 76% of identified mutations. ARID1B and SMARCB1 mutations were also found in individuals with the initial diagnosis of NCBRS. These individuals apparently belong to a small subset who display an intermediate CSS/NCBRS phenotype. Our proposed genotype-phenotype correlations are important for molecular screening strategies.

Group testing with DNA chips: generating designs and decoding experiments

DNA microarrays are a valuable tool for massively parallel DNA-DNA hybridization experiments. Currently, most applications rely on the existence of sequence-specific oligonucleotide probes. In large families of closely related target sequences, such as different virus subtypes, the high degree of similarity often makes it impossible to find a unique probe for every target. Fortunately, this is unnecessary. We propose a microarray design methodology based on a group testing approach. While probes might bind to multiple targets simultaneously, a properly chosen probe set can still unambiguously distinguish the presence of one target set from the presence of a different target set. Our method is the first one that explicitly takes cross-hybridization and experimental errors into account while accommodating several targets. The approach consists of three steps: (1) Pre-selection of probe candidates, (2) Generation of a suitable group testing design, and (3) Decoding of hybridization results to infer presence or absence of individual targets. Our results show that this approach is very promising, even for challenging data sets and experimental error rates of up to 5%. On a data set of 28S rDNA sequences we were able to identify 660 sequences, a substantial improvement over a prior approach using unique probes which only identified 408 sequences.

Large scale clustering of protein sequences with FORCE -A layout based heuristic for weighted cluster editing

BackgroundDetecting groups of functionally related proteins from their amino acid sequence alone has been a long-standing challenge in computational genome research. Several clustering approaches, following different strategies, have been published to attack this problem. Today, new sequencing technologies provide huge amounts of sequence data that has to be efficiently clustered with constant or increased accuracy, at increased speed.ResultsWe advocate that the model of weighted cluster editing, also known as transitive graph projection is well-suited to protein clustering. We present the FORCE heuristic that is based on transitive graph projection and clusters arbitrary sets of objects, given pairwise similarity measures. In particular, we apply FORCE to the problem of protein clustering and show that it outperforms the most popular existing clustering tools (Spectral clustering, TribeMCL, GeneRAGE, Hierarchical clustering, and Affinity Propagation). Furthermore, we show that FORCE is able to handle huge datasets by calculating clusters for all 192 187 prokaryotic protein sequences (66 organisms) obtained from the COG database. Finally, FORCE is integrated into the corynebacterial reference database CoryneRegNet.ConclusionFORCE is an applicable alternative to existing clustering algorithms. Its theoretical foundation, weighted cluster editing, can outperform other clustering paradigms on protein homology clustering. FORCE is open source and implemented in Java. The software, including the source code, the clustering results for COG and CoryneRegNet, and all evaluation datasets are available at http://gi.cebitec.uni-bielefeld.de/comet/force/.