Prediction uncertainty estimates elucidate the limitation of current NSCLC subtype classification in representing mutational heterogeneity.

The heterogeneous pathogenesis and treatment response of non-small cell lung cancer (NSCLC) has led clinical treatment decisions to be guided by NSCLC subtypes, with lung adenocarcinoma and lung squamous cell carcinoma being the most common subtypes. While histology-based subtyping remains challenging, NSCLC subtypes were found to be distinct at the transcriptomic level. However, unlike genomic alterations, gene expression is generally not assessed in clinical routine. Since subtyping of NSCLC has remained elusive using mutational data, we aimed at developing a neural network model that simultaneously learns from adenocarcinoma and squamous cell carcinoma samples of other tissue types and is regularized using a neural network model trained from gene expression data. While substructures of the expression-based manifold were captured in the mutation-based manifold, NSCLC classification accuracy did not significantly improve. However, performance was increased when rejecting inconclusive samples using an ensemble-based approach capturing prediction uncertainty. Importantly, SHAP analysis of misclassified samples identified co-occurring mutations indicative of both NSCLC subtypes, questioning the current NSCLC subtype classification to adequately represent inherent mutational heterogeneity. Since our model captures mutational patterns linked to clinical heterogeneity, we anticipate it to be suited as foundational model of genomic data for clinically relevant prognostic or predictive downstream tasks.

Phosphorylation of the Bruchpilot N-terminus in Drosophila unlocks axonal transport of active zone building blocks.

Protein scaffolds at presynaptic active zone membranes control information transfer at synapses. For scaffold biogenesis and maintenance, scaffold components must be safely transported along axons. A spectrum of kinases has been suggested to control transport of scaffold components, but direct kinase-substrate relationships and operational principles steering phosphorylation-dependent active zone protein transport are presently unknown. Here, we show that extensive phosphorylation of a 150-residue unstructured region at the N-terminus of the highly elongated Bruchpilot (BRP) active zone protein is crucial for ordered active zone precursor transport in Drosophila Point mutations that block SRPK79D kinase-mediated phosphorylation of the BRP N-terminus interfered with axonal transport, leading to BRP-positive axonal aggregates that also contain additional active zone scaffold proteins. Axonal aggregates formed only in the presence of non-phosphorylatable BRP isoforms containing the SRPK79D-targeted N-terminal stretch. We assume that specific active zone proteins are pre-assembled in transport packages and are thus co-transported as functional scaffold building blocks. Our results suggest that transient post-translational modification of a discrete unstructured domain of the master scaffold component BRP blocks oligomerization of these building blocks during their long-range transport.

WIsH: who is the host? Predicting prokaryotic hosts from metagenomic phage contigs.

SUMMARY: WIsH predicts prokaryotic hosts of phages from their genomic sequences. It achieves 63% mean accuracy when predicting the host genus among 20 genera for 3 kbp-long phage contigs. Over the best current tool, WisH shows much improved accuracy on phage sequences of a few kbp length and runs hundreds of times faster, making it suited for metagenomics studies. AVAILABILITY AND IMPLEMENTATION: OpenMP-parallelized GPL-licensed C ++ code available at https://github.com/soedinglab/wish. CONTACT: clovis.galiez@mpibpc.mpg.de or soeding@mpibpc.mpg.de. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Elastography Complements Ultrasound as Principle Modality in Breast Lesion Assessment.

BACKGROUND: The study aimed to improve breast cancer diagnosis with new ultrasound (US) modalities. We examined whether real-time elastography (RTE) complements the diagnostic performance of US. METHODS: The Ethical Committee approved the study. Patients provided written informed consent and received a whole breast workup. Breast lesions were evaluated by US and RTE. Lesions were assessed by BI-RADS, Tsukuba score (TS) and strain ratio (SR). RESULTS: The study included 164 breast lesions of which 101 were benign and 63 malignant. Women with benign lesions were predominantly premenopausal with sonographic dense breast tissue. Women with breast cancer were mostly postmenopausal, with a low sonographic density. Benign lesions had a mean TS of 2.05, which was significantly lower than the mean TS of 3.25 for malignant lesions. The SR for benign lesions (SR 1.83) was significantly smaller than for breast cancer (SR 4.83). Sensitivity and specificity was 95 and 81% for BI-RADS, 39 and 94% for TS, and 57 and 83% for SR, with a cutoff at 2.5. The combination of BI-RADS, TS and SR yielded a sensitivity of 95% and a specificity of 85%. CONCLUSION: Besides morphologic features revealed by US, elastic properties of breast lesions obtained by RTE can be exploited for diagnostic breast imaging.

Inter- and Intra-Observer Agreement in Ultrasound BI-RADS Classification and Real-Time Elastography Tsukuba Score Assessment of Breast Lesions.

Our aim was to prospectively evaluate inter- and intra-observer agreement between Breast Imaging Reporting and Data System (BI-RADS) classifications and Tsukuba elasticity scores (TSs) of breast lesions. The study included 164 breast lesions (63 malignant, 101 benign). The BI-RADS classification and TS of each breast lesion was assessed by the examiner and twice by three reviewers at an interval of 2 months. Weighted κ values for inter-observer agreement ranged from moderate to substantial for BI-RADS classification (κ = 0.585-0.738) and was substantial for TS (κ = 0.608-0.779). Intra-observer agreement was almost perfect for ultrasound (US) BI-RADS (κ = 0.847-0.872) and TS (κ = 0.879-0.914). Overall, individual reviewers are highly self-consistent (almost perfect intra-observer agreement) with respect to BI-RADS classification and TS, whereas inter-observer agreement was moderate to substantial. Comprehensive training is essential for achieving high agreement and minimizing the impact of subjectivity. Our results indicate that breast US and real-time elastography can achieve high diagnostic performance.

Bayesian Markov models consistently outperform PWMs at predicting motifs in nucleotide sequences.

Position weight matrices (PWMs) are the standard model for DNA and RNA regulatory motifs. In PWMs nucleotide probabilities are independent of nucleotides at other positions. Models that account for dependencies need many parameters and are prone to overfitting. We have developed a Bayesian approach for motif discovery using Markov models in which conditional probabilities of order k - 1 act as priors for those of order k This Bayesian Markov model (BaMM) training automatically adapts model complexity to the amount of available data. We also derive an EM algorithm for de-novo discovery of enriched motifs. For transcription factor binding, BaMMs achieve significantly (P = 1/16) higher cross-validated partial AUC than PWMs in 97% of 446 ChIP-seq ENCODE datasets and improve performance by 36% on average. BaMMs also learn complex multipartite motifs, improving predictions of transcription start sites, polyadenylation sites, bacterial pause sites, and RNA binding sites by 26-101%. BaMMs never performed worse than PWMs. These robust improvements argue in favour of generally replacing PWMs by BaMMs.

A high affinity RIM-binding protein/Aplip1 interaction prevents the formation of ectopic axonal active zones.

Synaptic vesicles (SVs) fuse at active zones (AZs) covered by a protein scaffold, at Drosophila synapses comprised of ELKS family member Bruchpilot (BRP) and RIM-binding protein (RBP). We here demonstrate axonal co-transport of BRP and RBP using intravital live imaging, with both proteins co-accumulating in axonal aggregates of several transport mutants. RBP, via its C-terminal Src-homology 3 (SH3) domains, binds Aplip1/JIP1, a transport adaptor involved in kinesin-dependent SV transport. We show in atomic detail that RBP C-terminal SH3 domains bind a proline-rich (PxxP) motif of Aplip1/JIP1 with submicromolar affinity. Pointmutating this PxxP motif provoked formation of ectopic AZ-like structures at axonal membranes. Direct interactions between AZ proteins and transport adaptors seem to provide complex avidity and shield synaptic interaction surfaces of pre-assembled scaffold protein transport complexes, thus, favouring physiological synaptic AZ assembly over premature assembly at axonal membranes.

Recruitment of TREX to the transcription machinery by its direct binding to the phospho-CTD of RNA polymerase II.

Messenger RNA (mRNA) synthesis and export are tightly linked, but the molecular mechanisms of this coupling are largely unknown. In Saccharomyces cerevisiae, the conserved TREX complex couples transcription to mRNA export and mediates mRNP formation. Here, we show that TREX is recruited to the transcription machinery by direct interaction of its subcomplex THO with the serine 2-serine 5 (S2/S5) diphosphorylated CTD of RNA polymerase II. S2 and/or tyrosine 1 (Y1) phosphorylation of the CTD is required for TREX occupancy in vivo, establishing a second interaction platform necessary for TREX recruitment in addition to RNA. Genome-wide analyses show that the occupancy of THO and the TREX components Sub2 and Yra1 increases from the 5' to the 3' end of the gene in accordance with the CTD S2 phosphorylation pattern. Importantly, in a mutant strain, in which TREX is recruited to genes but does not increase towards the 3' end, the expression of long transcripts is specifically impaired. Thus, we show for the first time that a 5'-3' increase of a protein complex is essential for correct expression of the genome. In summary, we provide insight into how the phospho-code of the CTD directs mRNP formation and export through TREX recruitment.

The Bruchpilot cytomatrix determines the size of the readily releasable pool of synaptic vesicles.

Synaptic vesicles (SVs) fuse at a specialized membrane domain called the active zone (AZ), covered by a conserved cytomatrix. How exactly cytomatrix components intersect with SV release remains insufficiently understood. We showed previously that loss of the Drosophila melanogaster ELKS family protein Bruchpilot (BRP) eliminates the cytomatrix (T bar) and declusters Ca(2+) channels. In this paper, we explored additional functions of the cytomatrix, starting with the biochemical identification of two BRP isoforms. Both isoforms alternated in a circular array and were important for proper T-bar formation. Basal transmission was decreased in isoform-specific mutants, which we attributed to a reduction in the size of the readily releasable pool (RRP) of SVs. We also found a corresponding reduction in the number of SVs docked close to the remaining cytomatrix. We propose that the macromolecular architecture created by the alternating pattern of the BRP isoforms determines the number of Ca(2+) channel-coupled SV release slots available per AZ and thereby sets the size of the RRP.

P-value-based regulatory motif discovery using positional weight matrices.

To analyze gene regulatory networks, the sequence-dependent DNA/RNA binding affinities of proteins and noncoding RNAs are crucial. Often, these are deduced from sets of sequences enriched in factor binding sites. Two classes of computational approaches exist. The first describe binding motifs by sequence patterns and search the patterns with highest statistical significance for enrichment. The second class uses the more powerful position weight matrices (PWMs). Instead of maximizing the statistical significance of enrichment, they maximize a likelihood. Here we present XXmotif (eXhaustive evaluation of matriX motifs), the first PWM-based motif discovery method that can optimize PWMs by directly minimizing their P-values of enrichment. Optimization requires computing millions of enrichment P-values for thousands of PWMs. For a given PWM, the enrichment P-value is calculated efficiently from the match P-values of all possible motif placements in the input sequences using order statistics. The approach can naturally combine P-values for motif enrichment, conservation, and localization. On ChIP-chip/seq, miRNA knock-down, and coexpression data sets from yeast and metazoans, XXmotif outperformed state-of-the-art tools, both in numbers of correctly identified motifs and in the quality of PWMs. In segmentation modules of D. melanogaster, we detect the known key regulators and several new motifs. In human core promoters, XXmotif reports most previously described and eight novel motifs sharply peaked around the transcription start site, among them an Initiator motif similar to the fly and yeast versions. XXmotif's sensitivity, reliability, and usability will help to leverage the quickly accumulating wealth of functional genomics data.

Chromatin modification of Notch targets in olfactory receptor neuron diversification.

Neuronal-class diversification is central during neurogenesis. This requirement is exemplified in the olfactory system, which utilizes a large array of olfactory receptor neuron (ORN) classes. We discovered an epigenetic mechanism in which neuron diversity is maximized via locus-specific chromatin modifications that generate context-dependent responses from a single, generally used intracellular signal. Each ORN in Drosophila acquires one of three basic identities defined by the compound outcome of three iterated Notch signaling events during neurogenesis. Hamlet, the Drosophila Evi1 and Prdm16 proto-oncogene homolog, modifies cellular responses to these iteratively used Notch signals in a context-dependent manner, and controls odorant receptor gene choice and ORN axon targeting specificity. In nascent ORNs, Hamlet erases the Notch state inherited from the parental cell, enabling a modified response in a subsequent round of Notch signaling. Hamlet directs locus-specific modifications of histone methylation and histone density and controls accessibility of the DNA-binding protein Suppressor of Hairless at the Notch target promoter.

RIM-binding protein, a central part of the active zone, is essential for neurotransmitter release.

The molecular machinery mediating the fusion of synaptic vesicles (SVs) at presynaptic active zone (AZ) membranes has been studied in detail, and several essential components have been identified. AZ-associated protein scaffolds are viewed as only modulatory for transmission. We discovered that Drosophila Rab3-interacting molecule (RIM)-binding protein (DRBP) is essential not only for the integrity of the AZ scaffold but also for exocytotic neurotransmitter release. Two-color stimulated emission depletion microscopy showed that DRBP surrounds the central Ca(2+) channel field. In drbp mutants, Ca(2+) channel clustering and Ca(2+) influx were impaired, and synaptic release probability was drastically reduced. Our data identify RBP family proteins as prime effectors of the AZ scaffold that are essential for the coupling of SVs, Ca(2+) channels, and the SV fusion machinery.

Uniform transitions of the general RNA polymerase II transcription complex.

We present genome-wide occupancy profiles for RNA polymerase (Pol) II, its phosphorylated forms and transcription factors in proliferating yeast. Pol II exchanges initiation factors for elongation factors during a 5' transition that is completed 150 nucleotides downstream of the transcription start site (TSS). The resulting elongation complex is composed of all the elongation factors and shows high levels of Ser7 and Ser5 phosphorylation on the C-terminal repeat domain (CTD) of Pol II. Ser2 phosphorylation levels increase until 600-1,000 nucleotides downstream of the TSS and do not correlate with recruitment of Spt6 and Pcf11, which bind the Ser2-phosphorylated CTD in vitro. This indicates CTD-independent recruitment mechanisms and CTD masking in vivo. Elongation complexes are productive and disassemble in a two-step 3' transition. Paf1, Spt16 (part of the FACT complex), and the CTD kinases Bur1 and Ctk1 exit upstream of the polyadenylation site, whereas Spt4, Spt5, Spt6, Spn1 (also called Iws1) and Elf1 exit downstream. Transitions are uniform and independent of gene length, type and expression.

Drosophila motor axons recognize and follow a Sidestep-labeled substrate pathway to reach their target fields.

During development of the Drosophila nervous system, migrating motor axons contact and interact with different cell types before reaching their peripheral muscle fields. The axonal attractant Sidestep (Side) is expressed in most of these intermediate targets. Here, we show that motor axons recognize and follow Side-expressing cell surfaces from the ventral nerve cord to their target region. Contact of motor axons with Side-expressing cells induces the down-regulation of Side. In the absence of Side, the interaction with intermediate targets is lost. Misexpression of Side in side mutants strongly attracts motor axons to ectopic sites. We provide evidence that, on motor axons, Beaten path Ia (Beat) functions as a receptor or part of a receptor complex for Side. In beat mutants, motor axons no longer recognize Side-expressing cell surfaces. Furthermore, Beat interacts with Side both genetically and biochemically. These results suggest that the tracing of Side-labeled cell surfaces by Beat-expressing growth cones is a major principle of motor axon guidance in Drosophila.