Michal Linial

Using Bayesian networks to analyze expression data

DNA hybridization arrays simultaneously measure the expression level for thousands of genes. These measurements provide a “snapshot” of transcription levels within the cell. A major challenge in computational biology is to uncover, from such measurements, gene/protein interactions and key biological features of cellular systems. In this paper, we propose a new framework for discovering interactions between genes based on multiple expression measurements This framework builds on the use of Bayesian networks for representing statistical dependencies. A Bayesian network is a graph-based model of joint multi-variate probability distributions that captures properties of conditional independence between variables. Such models are attractive for their ability to describe complex stochastic processes, and for providing clear methodologies for learning from (noisy) observations. We start by showing how Bayesian networks can describe interactions between genes. We then present an efficient algorithm capable of learning such networks and statistical method to assess our confidence in their features. Finally, we apply this method to the S. cerevisiae cell-cycle measurements of Spellman et al. [35] to uncover biological features

Protein function annotation by homology-based inference.

With many genomes now sequenced, computational annotation methods to characterize genes and proteins from their sequence are increasingly important. The BioSapiens Network has developed tools to address all stages of this process, and here we review progress in the automated prediction of protein function based on protein sequence and structure.

ProtoMap: automatic classification of protein sequences, a hierarchy of protein families, and local maps of the protein space.

We investigate the space of all protein sequences in search of clusters of related proteins. Our aim is to automatically detect these sets, and thus obtain a classification of all protein sequences. Our analysis, which uses standard measures of sequence similarity as applied to an all‐vs.‐all comparison of SWISSPROT, gives a very conservative initial classification based on the highest scoring pairs. The many classes in this classification correspond to protein subfamilies. Subsequently we merge the subclasses using the weaker pairs in a two‐phase clustering algorithm. The algorithm makes use of transitivity to identify homologous proteins; however, transitivity is applied restrictively in an attempt to prevent unrelated proteins from clustering together. This process is repeated at varying levels of statistical significance. Consequently, a hierarchical organization of all proteins is obtained.

Viral adaptation to host: a proteome-based analysis of codon usage and amino acid preferences

Viruses differ markedly in their specificity toward host organisms. Here, we test the level of general sequence adaptation that viruses display toward their hosts. We compiled a representative data set of viruses that infect hosts ranging from bacteria to humans. We consider their respective amino acid and codon usages and compare them among the viruses and their hosts. We show that bacteria‐infecting viruses are strongly adapted to their specific hosts, but that they differ from other unrelated bacterial hosts. Viruses that infect humans, but not those that infect other mammals or aves, show a strong resemblance to most mammalian and avian hosts, in terms of both amino acid and codon preferences. In groups of viruses that infect humans or other mammals, the highest observed level of adaptation of viral proteins to host codon usages is for those proteins that appear abundantly in the virion. In contrast, proteins that are known to participate in host‐specific recognition do not necessarily adapt to their respective hosts. The implication for the potential of viral infectivity is discussed.

Efficient algorithms for accurate hierarchical clustering of huge datasets

Motivation: UPGMA (average linking) is probably the most popular algorithm for hierarchical data clustering, especially in computational biology. However, UPGMA requires the entire dissimilarity matrix in memory. Due to this prohibitive requirement, UPGMA is not scalable to very large datasets. Application: We present a novel class of memory-constrained UPGMA (MC-UPGMA) algorithms. Given any practical memory size constraint, this framework guarantees the correct clustering solution without explicitly requiring all dissimilarities in memory. The algorithms are general and are applicable to any dataset. We present a data-dependent characterization of hardness and clustering efficiency. The presented concepts are applicable to any agglomerative clustering formulation. Results: We apply our algorithm to the entire collection of protein sequences, to automatically build a comprehensive evolutionary-driven hierarchy of proteins from sequence alone. The newly created tree captures protein families better than state-of-the-art large-scale methods such as CluSTr, ProtoNet4 or single-linkage clustering. We demonstrate that leveraging the entire mass embodied in all sequence similarities allows to significantly improve on current protein family clusterings which are unable to directly tackle the sheer mass of this data. Furthermore, we argue that non-metric constraints are an inherent complexity of the sequence space and should not be overlooked. The robustness of UPGMA allows significant improvement, especially for multidomain proteins, and for large or divergent families. Availability: A comprehensive tree built from all UniProt sequence similarities, together with navigation and classification tools will be made available as part of the ProtoNet service. A C++ implementation of the algorithm is available on request. Contact: lonshy@cs.huji.ac.il

Voltage-dependent interaction between the muscarinic ACh receptor and proteins of the exocytic machinery.

1 Release of neurotransmitter into the synaptic cleft is the last step in the chain of molecular events following the arrival of an action potential at the nerve terminal. The neurotransmitter exerts negative feedback on its own release. This inhibition would be most effective if exerted on the first step in this chain of events, i.e. a step that is mediated by membrane depolarization. Indeed, in numerous studies feedback inhibition was found to be voltage dependent. 2 The purpose of this study is to investigate whether the mechanism underlying feedback inhibition of transmitter release resides in interaction between the presynaptic autoreceptors and the exocytic apparatus, specifically the soluble NSF‐attachment protein receptor (SNARE) complex. 3 Using rat synaptosomes we show that the muscarinic ACh autoreceptor (mAChR) is an integral component of the exocytic machinery. It interacts with syntaxin, synaptosomal‐associated protein of 25 kDa (SNAP‐25), vesicle‐associated membrane protein (VAMP) and synaptotagmin as shown using both cross‐linking and immunoprecipitation. 4 The interaction between mAChRs and both syntaxin and SNAP‐25 is modulated by depolarization levels; binding is maximal at resting potential and disassembly occurs at higher depolarization. 5 This voltage‐dependent interaction of mAChRs with the secretory core complex appears suitable for controlling the rapid, synchronous neurotransmitter release at nerve terminals.

SNARE Proteins-Why So Many, Why So Few?

Abstract: Both trafficking and secretion critically depend on accurate and specific membrane recognition and fusion. A key step in these processes is the assembly of a complex consisting of a small number of proteins, i.e., the exocytic core complex. In nerve terminals, this set consists of VAMP and synaptotagmin, which reside at membranes of synaptic vesicles, and syntaxin and SNAP‐25 at the plasma membrane. In this survey, different secretory systems that depend on the exocytic core proteins are considered. The possibility that specificity in membrane recognition and fusion is achieved by the numerous variants of proteins of the exocytic core is discussed. Variability of the core complex proteins is determined by the complexity of gene families, isoform‐specific localization, and posttranslational modifications. Basic biochemical properties depend on specific isoforms, and the possible protein‐protein interactions are determined, in turn, by the compatibility of different isoforms. A correlation between specific variants and distinct biochemical or cellular properties is shown. The outcome of this survey is that heterogeneity in secretion may be dictated by the large number of possible combinations of variants of only a few proteins.

Nucleotide bindind by the synapse associated protein SAP90

The rat synapse associated protein SAP90 is a member of a superfamily of potential guanylate kinases localized at cell‐cell contact sites. This superfamily includes the synapse associated protein SAP97, a close relative of SAP90, the Drosophila tumor suppressor gene product dlg‐Ap, the mammalian zonula occludens proteins ZO‐1 and ZO‐2 and the erythrocyte protein p55. Here we show that SAP90 specifically binds GMP in the micromolar range while binding to ATP, GDP and ADP is at a much lower affinity (10–25 mM), whether or not binding is detected for other guanine and adenine nucleotides. No guanylate kinase activity of SAP90 was detected under our experimental conditions. The importance of the GMP binding capacity per se and an evolutionary role for conserving of the guanylate kinase domain in this superfamily are discussed.

ProtoNet 4.0: A hierarchical classification of one million protein sequences

ProtoNet is an automatic hierarchical classification of the protein sequence space. In 2004, the ProtoNet (version 4.0) presents the analysis of over one million proteins merged from SwissProt and TrEMBL databases. In addition to rich visualization and analysis tools to navigate the clustering hierarchy, we incorporated several improvements that allow a simplified view of the scaffold of the proteins. An unsupervised, biologically valid method that was developed resulted in a condensation of the ProtoNet hierarchy to only 12% of the clusters. A large portion of these clusters was automatically assigned high confidence biological names according to their correspondence with functional annotations. ProtoNet is available at: http://www.protonet.cs.huji.ac.il.

Anxiety induced by prenatal stress is associated with suppression of hippocampal genes involved in synaptic function.

Exposure of pregnant women or animals to stress during a critical period of foetal brain development increases the likelihood of anxiety, depression and learning deficits that are associated with structural alterations in the offspring hippocampus. In this study, we report the effect of gestational stress in rats on anxiogenic behaviour and hippocampal gene expression of their 23‐day‐old female offspring. As the rat brain continues to develop after birth, we also used the procedure of handling (H) during the first 10 days of life to reverse the anxiogenic behaviour of prenatally stressed (PS) rats. By means of micro‐array analysis on hippocampal extracts, we found that the expression of about 6.1% of 9505 valid genes was significantly altered by prenatal stress (p < 0.05). Of these, 48% were over‐expressed and 52% under‐expressed. The latter included ∼300 genes that participate in axonal growth, regulation of ion channels and transporters, trafficking of synaptic vesicles and neurotransmitter release. About 30% of the genes that were down‐regulated in PS rats were restored to control levels by H. These include genes that play a role in pre‐synaptic organization and function. Our results provide a possible relationship between hippocampal gene expression and changes in behaviour resulting from prenatal stress.