Similar Phylogenetic Profiles Research Articles

Abstract A46: Charting the synthetic lethality landscape in pediatric cancer to advance whole-exome precision-based treatments

Abstract One of the challenges in pediatric cancer (PC) research is that cancers in children are uncommon and are quite different from adults. Much of the research in adult cancers is focused on studying cancer driver genes, aiming at their therapeutic targeting. However, PCs are often driven by relatively few genetic alterations that are distinct from those that occur in adult cancers. Here we apply a novel data-driven approach to identify the synthetic lethality (SL) networks of several different pediatric cancers. These provide a new platform for discovering novel vulnerabilities in primary tumors from PCs that extends previous approaches commonly used in adult cancer. SL interactions denote the relationship between two genes whose combined inactivation is lethal to the cell, while their individual inactivation is not. To identify the SL landscape characteristic of a specific pediatric cancer, we mined the relevant pediatric cell line and patients’ tumor data in the TARGET database. Our computational framework consists of four inference steps: For SL interactions, we first identify putative SL gene pairs from the pediatric cell line dependency map generated by in vitro RNAi/CRISPR screens (depmap). Second, among the candidate gene pairs that pass the first step, we select those gene pairs whose co-inactivation is under-represented in pediatric tumors, indicating that they are selected against. Third, we further prioritize candidate SL pairs whose co-inactivation is associated with better prognosis, indicating that they may hamper tumor progression. Finally, we prioritize SL paired genes with similar evolutionary phylogenetic profiles. Applying this approach to analyze TARGET data, we identify the first genome-wide SL networks in five pediatric tumors including Wilms’ tumor, neuroblastoma, AML, ALL, and osteosarcoma. The predicted SL interactions are first tested and validated via experimental in vitro CRISPR screens. Second, we show that the PC specific SL networks are predictive of drug response in pediatric cell lines but not in adult cell lines of the corresponding tumor type. These results establish that the predicted SL interactions offer an exciting venue for developing predictive biomarkers specific for PC treatments. Importantly, these predictions were performed in an unsupervised manner, reducing the known risk of over-fitting and lack of generalizability commonly associated with supervised prediction methods. Notably, our analysis identifies many SL partners of key drivers of PCs such as WT1, MYCN, and ATRX, and the key interactions discovered include ATRX-MAP kinases, MYCN-CDC6 (cell cycle regulation), and DNMT1-HK2. These provide novel selective drug target candidates for the tumors driven by these genes and lay a basis for new treatment combinations. Taken together, these results lay a basis for a new paradigm for whole-exome SL-based precision treatments in pediatric oncology, complementing existing mutation- and fusion-based approaches. Citation Format: Fiorella Schischlik, Joo Sang Lee, Nirali Shah, Rosandra N. Kaplan, Carol J. Thiele, Brigitte Widemann, Eytan Ruppin. Charting the synthetic lethality landscape in pediatric cancer to advance whole-exome precision-based treatments [abstract]. In: Proceedings of the AACR Special Conference on the Advances in Pediatric Cancer Research; 2019 Sep 17-20; Montreal, QC, Canada. Philadelphia (PA): AACR; Cancer Res 2020;80(14 Suppl):Abstract nr A46.

Duckweed hosts a taxonomically similar bacterial assemblage as the terrestrial leaf microbiome.

Culture-independent characterization of microbial communities associated with popular plant model systems have increased our understanding of the plant microbiome. However, the integration of other model systems, such as duckweed, could facilitate our understanding of plant microbiota assembly and evolution. Duckweeds are floating aquatic plants with many characteristics, including small size and reduced plant architecture, that suggest their use as a facile model system for plant microbiome studies. Here, we investigated the structure and assembly of the duckweed bacterial microbiome. First, a culture-independent survey of the duckweed bacterial microbiome from different locations in New Jersey revealed similar phylogenetic profiles. These studies showed that Proteobacteria is a dominant phylum in the duckweed bacterial microbiome. To observe the assembly dynamics of the duckweed bacterial community, we inoculated quasi-gnotobiotic duckweed with wastewater effluent from a municipal wastewater treatment plant. Our results revealed that duckweed strongly shapes its bacterial microbiome and forms distinct associations with bacterial community members from the initial inoculum. Additionally, these inoculation studies showed the bacterial communities of different duckweed species were similar in taxa composition and abundance. Analysis across the different duckweed bacterial communities collected in this study identified a set of "core" bacterial taxa consistently present on duckweed irrespective of the locale and context. Furthermore, comparison of the duckweed bacterial community to that of rice and Arabidopsis revealed a conserved taxonomic structure between the duckweed microbiome and the terrestrial leaf microbiome. Our results suggest that duckweeds utilize similar bacterial community assembly principles as those found in terrestrial plants and indicate a highly conserved structuring effect of leaf tissue on the plant microbiome.

Abstract A188: Harnessing synthetic lethality to predict the response to cancer treatments

Abstract Synthetic lethality (SL) describes an interaction between a pair of genes whereby their double knockout is lethal, while their respective knockout is not. The identification of SL interactions (SLi) via large-scale genomic screens offers promising opportunities for developing selective therapies in cancer. However, our analysis of the TCGA cohort shows that many of the interactions do not carry predictive signal of patient survival or drug response. Here we present a data-driven approach termed ISLE (Identification of clinically relevant Synthetic LEthality) that mines the TCGA cohort to identify a subset of clinically relevant SL interactions (cSLi). ISLE consists of the following inference steps, analysis of tumor, cell line, and gene evolutionary data. We first create an initial pool of SL pairs identified through direct double knockout screens/isogenic cell line screens or inferred from large-scale shRNA/sgRNA single-gene knockout screens. Starting from this initial SL pool, ISLE first identifies putative SL gene pairs whose co-inactivation is under-represented in tumors, testifying that it is selected against. Second, it prioritizes candidate SL pairs whose co-inactivation is associated with improved patient’s prognosis, testifying that it may hamper tumor progression. Finally, it prioritizes SL-gene pairs with similar evolutionary phylogenetic profiles based on the notion that SL interactions are conserved across multiple species. We validate the identified SL pairs using an unseen large-scale in vitro drug response screen by showing the SL pairs marks a decent prediction accuracy (AUC~0.8). We compare ISLE’s performance to the standard supervised drug response prediction approaches in DREAM challenges, and our prediction based on generic pretreatment tumor samples (from TCGA) was within top 3 in prediction accuracy among the top predictors. ISLE-based approach also successfully distinguishes responders vs nonresponders to drug treatment (for >70% of drugs) in mouse xenografts using the activity profile of the drug target’s SL-partners. We then experimentally show the utility of SL in predicting synergistic drug combinations in patient-derived cell lines based on the notion that the two drugs whose targets have SL interactions are synergistic. Most importantly, we demonstrate for the first time that an SL network can successfully predict the treatment outcome in cancer patients in multiple large-scale patient datasets including TCGA, where cSLi are successfully predict patients’ response for more than 70% of cancer drugs. ISLE is predictive of patients’ response for the majority of current cancer drugs without any drug-specific training. Of paramount importance, the predictions of ISLE are based on SLi between (potentially) all genes in the cancer genome, thus prioritizing treatments for patients whose tumors do not bear specific actionable mutations in cancer driver genes, offering a novel approach to precision-based cancer therapy. Citation Format: Joo S. Lee, Avinash Das, Livnat Jerby-Arnon, Rand Arafeh, Matthew Davidson, Arnaud Amzallag, Seung Gu Park, Kuoyuan Cheng, Welles Robinson, Dikla Atias, Chani Stossel, Ella Buzhor, Gidi Stein, Joshua J. Waterfall, Paul S. Meltzer, Talia Golan, Sridhar Hannenhalli, Eyal Gottlieb, Cyril H. Benes, Yardena Samuels, Emma Shanks, Eytan Ruppin. Harnessing synthetic lethality to predict the response to cancer treatments [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr A188.

Abstract PR09: Harnessing synthetic lethality to predict clinical outcomes of cancer treatment

Abstract Significance: The identification of Synthetic Lethal interactions (SLi) have long been considered a foundation for the advancement of cancer treatment. The rapidly accumulating large-scale patient data now provides a golden opportunity to infer SLi directly from patient samples. Here we present a new data-driven approach termed ISLE for identifying SLi, which is then shown to be predictive of clinical outcomes of cancer treatment in an unsupervised manner, for the first time. Methods: ISLE consists of four inference steps, analyzing tumor, cell line and gene evolutionary data: It first identifies putative SL gene pairs whose co-inactivation is underrepresented in tumors, testifying that they are selected against. Second, it further prioritizes candidate SL pairs whose co-inactivation is associated with better prognosis in patients, testifying that they may hamper tumor progression. Finally, it eliminates false positive SLi using gene essentiality screens (testifying to causal SLi relations) and prioritizing SLi paired genes with similar evolutionary phylogenetic profiles. Results: We applied ISLE to analyze the TCGA tumor collection and generated the first clinically-derived pan-cancer SL-network, composed of SLi common across many cancer types. We validated that these SLi match the known, experimentally identified SLi (AUC=0.87), and show that the SL-network is predictive of patient survival in an independent breast cancer dataset (METABRIC). Based on the predicted SLi, we predicted drug response in a wide variety of in vitro, mouse xenograft and patient data, altogether encompassing >700 single drugs and >5,000 drug combinations in >1,000 cell lines, 375 xenograft models and >5,000 patient samples. Importantly, these predictions were performed in an unsupervised manner, reducing the known risk of over-fitting the data commonly associated with supervised prediction methods. SL-derived predictions are based on computing an SL-score that estimates the efficacy of a given drug in a given tumor based on the latter's omics data. The SL-score counts the number of inactive SL-partners of a given drug target(s) in the given tumor, reflecting the notion that a drug is likely to be more effective in tumors where many of its targets' SL-partners are inactive. The predicted SL-scores show significant correlations (R > 0.4) with large-scale in vitro and in vivo drug response screens for the majority of drugs tested. Based on the conjecture that synergism between drugs may be mediated by underlying SLi between their targets, we additionally provide accurate predictions of drug synergism for both in vitro and in vivo drug combination screens (AUC~0.8). Most importantly, we demonstrate for the first time that an SL-network can successfully predict the treatment outcome in cancer patients in multiple large-scale patient datasets including the TCGA, where SLis successfully predict patients' response for 75% of cancer drugs. Conclusions: ISLE is predictive of the patients' response for the majority of current cancer drugs. Of paramount importance, the predictions of ISLE are based on SLi between (potentially) all genes in the cancer genome, thus prioritizing treatments for patients whose tumors do not bear specific actionable mutations in cancer driver genes, offering a novel approach to precision-based cancer therapy. The predictive performance of ISLE is likely to further improve with the expected rapid accumulation of additional cancer omics and clinical phenotypic data. Citation Format: Joo Sang Lee, Avinash Das, Livnat Jerby-Arnon, Dikla Atias, Arnaud Amzallag, Cyril H. Benes, Talia Golan, Eytan Ruppin. Harnessing synthetic lethality to predict clinical outcomes of cancer treatment [abstract]. In: Proceedings of the AACR Precision Medicine Series: Opportunities and Challenges of Exploiting Synthetic Lethality in Cancer; Jan 4-7, 2017; San Diego, CA. Philadelphia (PA): AACR; Mol Cancer Ther 2017;16(10 Suppl):Abstract nr PR09.

Abstract 543: Harnessing synthetic lethality to predict clinical outcomes of cancer treatment

Abstract Significance: The identification of Synthetic Lethal interactions (SLi) has long been considered a foundation for the advancement of cancer treatment. The rapidly accumulating large-scale patient data now provides a golden opportunity to infer SLi directly from patient samples. Here we present a new data-driven approach termed ISLE for identifying SLi, which is then shown to be predictive of clinical outcomes of cancer treatment in an unsupervised manner, for the first time. Methods: ISLE consists of four inference steps, analyzing tumor, cell line and gene evolutionary data: It first identifies putative SL gene pairs whose co-inactivation is underrepresented in tumors, testifying that they are selected against. Second, it further prioritizes candidate SL pairs whose co-inactivation is associated with better prognosis in patients, testifying that they may hamper tumor progression. Finally, it eliminates false positive SLi using gene essentiality screens (testifying to causal SLi relations) and prioritizing SLi paired genes with similar evolutionary phylogenetic profiles. Results: We applied ISLE to analyze the TCGA tumor collection and generated the first clinically-derived pan-cancer SL-network, composed of SLi common across many cancer types. We validated that these SLi match the known, experimentally identified SLi (AUC=0.87), and show that the SL-network is predictive of patient survival in an independent breast cancer dataset (METABRIC). Based on the predicted SLi, we predicted drug response of single agents and drug combinations in a wide variety of in vitro, mouse xenograft and patient data, altogether encompassing >700 single drugs and >5,000 drug combinations in >1,000 cell lines, 375 xenograft models and >5,000 patient samples. Of note, these predictions were performed in an unsupervised manner, reducing the known risk of over-fitting the data commonly associated with supervised prediction methods. Our prediction is based on the notion that a drug is likely to be more effective in tumors where many of its targets’ SL-partners are inactive, and drug synergism may be mediated by underlying SLi between their targets. Most importantly, we demonstrate for the first time that an SL-network can successfully predict the treatment outcome in cancer patients in multiple large-scale patient datasets including the TCGA, where SLis successfully predict patients’ response for 75% of cancer drugs. Conclusions: ISLE is predictive of the patients’ response for the majority of current cancer drugs. Of paramount importance, the predictions of ISLE are based on SLi between (potentially) all genes in the cancer genome, thus prioritizing treatments for patients whose tumors do not bear specific actionable mutations in cancer driver genes, offering a novel approach to precision-based cancer therapy. The predictive performance of ISLE is likely to further improve with the expected rapid accumulation of additional patient data. Citation Format: Joo Sang Lee, Avinash Das, Livnat Jerby-Arnon, Seung Gu Park, Matthew Davidson, Dikla Atias, Arnaud Amzallag, Chani Stossel, Ella Buzhor, Welles Robinson, Kuoyuan Cheng, Joshua J. Waterfall, Paul S. Meltzer, Sridhar Hannenhalli, Cyril H. Benes, Talia Golan, Emma Shanks, Eytan Ruppin. Harnessing synthetic lethality to predict clinical outcomes of cancer treatment [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 543. doi:10.1158/1538-7445.AM2017-543

PhyloGene server for identification and visualization of co-evolving proteins using normalized phylogenetic profiles.

Proteins that function in the same pathways, protein complexes or the same environmental conditions can show similar patterns of sequence conservation across phylogenetic clades. In species that no longer require a specific protein complex or pathway, these proteins, as a group, tend to be lost or diverge. Analysis of the similarity in patterns of sequence conservation across a large set of eukaryotes can predict functional associations between different proteins, identify new pathway members and reveal the function of previously uncharacterized proteins. We used normalized phylogenetic profiling to predict protein function and identify new pathway members and disease genes. The phylogenetic profiles of tens of thousands conserved proteins in the human, mouse, Caenorhabditis elegans and Drosophila genomes can be queried on our new web server, PhyloGene. PhyloGene provides intuitive and user-friendly platform to query the patterns of conservation across 86 animal, fungal, plant and protist genomes. A protein query can be submitted either by selecting the name from whole-genome protein sets of the intensively studied species or by entering a protein sequence. The graphic output shows the profile of sequence conservation for the query and the most similar phylogenetic profiles for the proteins in the genome of choice. The user can also download this output in numerical form.

Human disease locus discovery and mapping to molecular pathways through phylogenetic profiling

Genes with common profiles of the presence and absence in disparate genomes tend to function in the same pathway. By mapping all human genes into about 1000 clusters of genes with similar patterns of conservation across eukaryotic phylogeny, we determined that sets of genes associated with particular diseases have similar phylogenetic profiles. By focusing on those human phylogenetic gene clusters that significantly overlap some of the thousands of human gene sets defined by their coexpression or annotation to pathways or other molecular attributes, we reveal the evolutionary map that connects molecular pathways and human diseases. The other genes in the phylogenetic clusters enriched for particular known disease genes or molecular pathways identify candidate genes for roles in those same disorders and pathways. Focusing on proteins coevolved with the microphthalmia-associated transcription factor (MITF), we identified the Notch pathway suppressor of hairless (RBP-Jk/SuH) transcription factor, and showed that RBP-Jk functions as an MITF cofactor.

Identification of small RNA pathway genes using patterns of phylogenetic conservation and divergence

Genetic and biochemical analyses of RNA interference (RNAi) and microRNA (miRNA) pathways have revealed proteins such as Argonaute/PIWI and Dicer that process and present small RNAs to their targets. Well validated small RNA pathway cofactors, such as the Argonaute/PIWI proteins show distinctive patterns of conservation or divergence in particular animal, plant, fungal, and protist species. We compared 86 divergent eukaryotic genome sequences to discern sets of proteins that show similar phylogenetic profiles with known small RNA cofactors. A large set of additional candidate small RNA cofactors have emerged from functional genomic screens for defects in miRNA- or siRNA-mediated repression in C. elegans and D. melanogaster1,2 and from proteomic analyses of proteins co-purifying with validated small RNA pathway proteins3,4. The phylogenetic profiles of many of these candidate small RNA pathway proteins are similar to those of known small RNA cofactor proteins. We used a Bayesian approach to integrate the phylogenetic profile analysis with predictions from diverse transcriptional coregulation and proteome interaction datasets to assign a probability for each protein for a role in a small RNA pathway. Testing high-confidence candidates from this analysis for defects in RNAi silencing, we found that about half of the predicted small RNA cofactors are required for RNAi silencing. Many of the newly identified small RNA pathway proteins are orthologues of proteins implicated in RNA splicing. In support of a deep connection between the mechanism of RNA splicing and small RNA-mediated gene silencing, the presence of the Argonaute proteins and other small RNA components in the many species analysed strongly correlates with the number of introns in that species.

Caryophyllales: Evaluating phylogenetic signal in trnK intron versus matK

Abstract We assess the phylogenetic information in trnK intron at the ordinal level using the Caryophyllales and compare it with that derived from matK. The trnK gene is split into two exons by an intron that includes the matK gene. The plastid trnK is a tRNA gene encoding Lysine(UUU), whereas the matK gene is a putative group II intron maturase. The two regions are usually coamplified, and trnK intron is partially sequenced but its sequences are often excluded from phylogenetic reconstruction at deep historic levels. This study shows that the two regions are comparable in proportion of variable sites, possess a comparable pattern of substitution rates per site, and display similar phylogenetic informativeness profiles and per‐site informativeness. Phylogenetic analyses show strong congruence between phylogenetic trees based on matK and trnK intron partitioned datasets from 45 genera representing 30 of the 34 recognized Caryophyllales families. The trnK intron alone provides a relatively well‐resolved topology for the order. Combining the trnK intron with matK sequence data resulted in six most parsimonious trees, differing only in the placement of Claytonia (Portulacaceae) within the noncore group. A well‐supported major basal split in the order into core and noncore Caryophyllales with Rhabdodendraceae, Simmondsiaceae, and Asteropeiaceae as sister to remaining core lineages is evident in partitioned and combined analyses. The placement of these three families has been disputable, impacting the overall backbone topology of the Caryophyllales. This study demonstrates the cost effectiveness of using the trnK intron along with matK (both substitutions and insertions/deletions) at deeper phylogenetic level.

A modular view of the bacteriophage genomic space: identification of host and lifestyle marker modules

Bacteriophage genomes can be regarded as an ensemble of modules which are accessible to the whole phage population via recombination. The time spent by prophages in the bacterial host provides them with the opportunity to exchange modules with other prophages or infecting phages. Here we analyze the modular structure of a set of 457 phages and 760 prophages extracted from completely sequenced bacterial genomes using the ACLAME database and its associated tools. We identified 91 modules of proteins with similar phylogenetic profiles. Of these, 25 and 6 are associated with temperate and virulent phages, respectively; 57 are restricted to a host or small group of hosts; and 55 could be annotated with a phage function. We use the transposable phages as a study case and show how the inclusion of prophages allows us to unveil new types of genome organization (i.e. novel module combinations) and obtain insight into the host range for this particular group, highlighting the utility of prophage prediction to better characterize phage diversity.

Identifying candidate cancer genes based on co-evolving gene modules

Data of somatic mutation screening of cancer genomes have provided us huge amounts of information for identifying new cancer genes. Current methods for identifying candidate cancer genes based on gene mutation frequencies tend to find cancer genes with high mutation frequencies. However, many genes with low mutation frequencies might also play important roles during tumorigenesis. Based on the assumption that genes with similar phylogenetic profiles and protein-protein interactions might have similar functions and their disruptions might lead to similar disease phenotypes, we proposed a new approach to find candidate cancer genes. First, we searched for protein-protein interaction subnetworks within which proteins have similar phylogenetic profiles, termed as co-evolving gene modules. Then, we identified genes that have at least one non-synonymous mutation in cancer genomes and directly interact with known cancer genes in the same co-evolving gene modules and predicted them as candidate cancer genes. In this way, we found 15 candidate cancer genes, among which only two genes had been identified previously as candidate cancer genes using the methods based on gene mutation frequencies. Thus, the candidate cancer genes with low mutation frequencies can be found by our method.

Testing the Accuracy of Eukaryotic Phylogenetic Profiles for Prediction of Biological Function

A phylogenetic profile captures the pattern of gene gain and loss throughout evolutionary time. Proteins that interact directly or indirectly within the cell to perform a biological function will often co-evolve, and this co-evolution should be well reflected within their phylogenetic profiles. Thus similar phylogenetic profiles are commonly used for grouping proteins into functional groups. However, it remains unclear how the size and content of the phylogenetic profile impacts the ability to predict function, particularly in Eukaryotes. Here we developed a straightforward approach to address this question by constructing a complete set of phylogenetic profiles for 31 fully sequenced Eukaryotes. Using Gene Ontology as our gold standard, we compared the accuracy of functional predictions made by a comprehensive array of permutations on the complete set of genomes. Our permutations showed that phylogenetic profiles containing between 25 and 31 Eukaryotic genomes performed equally well and significantly better than all other permuted genome sets, with one exception: we uncovered a core of group of 18 genomes that achieved statistically identical accuracy. This core group contained genomes from each branch of the eukaryotic phylogeny, but also contained several groups of closely related organisms, suggesting that a balance between phylogenetic breadth and depth may improve our ability to use Eukaryotic specific phylogenetic profiles for functional annotations.

The DART classification of unannotated transcription within the ENCODE regions: Associating transcription with known and novel loci

For the approximately 1% of the human genome in the ENCODE regions, only about half of the transcriptionally active regions (TARs) identified with tiling microarrays correspond to annotated exons. Here we categorize this large amount of "unannotated transcription." We use a number of disparate features to classify the 6988 novel TARs-array expression profiles across cell lines and conditions, sequence composition, phylogenetic profiles (presence/absence of syntenic conservation across 17 species), and locations relative to genes. In the classification, we first filter out TARs with unusual sequence composition and those likely resulting from cross-hybridization. We then associate some of those remaining with proximal exons having correlated expression profiles. Finally, we cluster unclassified TARs into putative novel loci, based on similar expression and phylogenetic profiles. To encapsulate our classification, we construct a Database of Active Regions and Tools (DART.gersteinlab.org). DART has special facilities for rapidly handling and comparing many sets of TARs and their heterogeneous features, synchronizing across builds, and interfacing with other resources. Overall, we find that approximately 14% of the novel TARs can be associated with known genes, while approximately 21% can be clustered into approximately 200 novel loci. We observe that TARs associated with genes are enriched in the potential to form structural RNAs and many novel TAR clusters are associated with nearby promoters. To benchmark our classification, we design a set of experiments for testing the connectivity of novel TARs. Overall, we find that 18 of the 46 connections tested validate by RT-PCR and four of five sequenced PCR products confirm connectivity unambiguously.

An improved method for identifying functionally linked proteins using phylogenetic profiles

BackgroundPhylogenetic profiles record the occurrence of homologs of genes across fully sequenced organisms. Proteins with similar profiles are typically components of protein complexes or metabolic pathways. Various existing methods measure similarity between two profiles and, hence, the likelihood that the two proteins co-evolve. Some methods ignore phylogenetic relationships between organisms while others account for such with metrics that explicitly model the likelihood of two proteins co-evolving on a tree. The latter methods more sensitively detect co-evolving proteins, but at a significant computational cost. Here we propose a novel heuristic to improve phylogenetic profile analysis that accounts for phylogenetic relationships between genomes in a computationally efficient fashion. We first order the genomes within profiles and then enumerate runs of consecutive matches and accurately compute the probability of observing these. We hypothesize that profiles with many runs are more likely to involve functionally related proteins than profiles in which all the matches are concentrated in one interval of the tree.ResultsWe compared our approach to various previously published methods that both ignore and incorporate the underlying phylogeny between organisms. To evaluate performance, we compare the functional similarity of rank-ordered lists of protein pairs that share similar phylogenetic profiles by assessing significance of overlap in their Gene Ontology annotations. Accounting for runs in phylogenetic profile matches improves our ability to identify functionally related pairs of proteins. Furthermore, the networks that result from our approach tend to have smaller clusters of co-evolving proteins than networks computed using previous approaches and are thus more useful for inferring functional relationships. Finally, we report that our approach is orders of magnitude more computationally efficient than full tree-based methods.ConclusionWe have developed an improved method for analyzing phylogenetic profiles. The method allows us to more accurately and efficiently infer functional relationships between proteins based on these profiles than other published approaches. As the number of fully sequenced genomes increases, it becomes more important to account for evolutionary relationships among organisms in comparative analyses. Our approach, therefore, serves as an important example of how these relationships may be accounted for in an efficient manner.

The distinctive signatures of promoter regions and operon junctions across prokaryotes

Here we show that regions upstream of first transcribed genes have oligonucleotide signatures that distinguish them from regions upstream of genes in the middle of operons. Databases of experimentally confirmed transcription units do not exist for most genomes. Thus, to expand the analyses into genomes with no experimentally confirmed data, we used genes conserved adjacent in evolutionarily distant genomes as representatives of genes inside operons. Likewise, we used divergently transcribed genes as representative examples of first transcribed genes. In model organisms, the trinucleotide signatures of regions upstream of these representative genes allow for operon predictions with accuracies close to those obtained with known operon data (0.8). Signature-based operon predictions have more similar phylogenetic profiles and higher proportions of genes in the same pathways than predicted transcription unit boundaries (TUBs). These results confirm that we are separating genes with related functions, as expected for operons, from genes not necessarily related, as expected for genes in different transcription units. We also test the quality of the predictions using microarray data in six genomes and show that the signature-predicted operons tend to have high correlations of expression. Oligonucleotide signatures should expand the number of tools available to identify operons even in poorly characterized genomes.

Inferring Functional Linkages between Proteins from Evolutionary Scenarios

Identifying potential protein interactions is of great importance in understanding the topologies of cellular networks, which is much needed and valued in current systematic biological studies. The development of our computational methods to predict protein-protein interactions have been spurred on by the massive sequencing efforts of the genomic revolution. Among these methods is phylogenetic profiling, which assumes that proteins under similar evolutionary pressures with similar phylogenetic profiles might be functionally related. Here, we introduce a method for inferring functional linkages between proteins from their evolutionary scenarios. The term evolutionary scenario refers to a series of events that occurred in speciation over time, which can be reconstructed given a phylogenetic profile and a species tree. Common evolutionary pressures on two proteins can then be inferred by comparing their evolutionary scenarios, which is a direct indication of their functional linkage. This scenario method has proven to have better performance compared with the classical phylogenetic profile method, when applied to the same test set. In addition, predicted results of the two methods are found to be fairly different, suggesting the possibility of merging them in order to achieve a better performance. We analyzed the influence of the topology of the phylogenetic tree on the performance of this method, and found it to be robust to perturbations in the topology of the tree. However, if a completely random tree is incorporated, performance will decline significantly. The evolutionary scenario method was used for inferring functional linkages in 67 species, and 40,006 linkages were predicted. We examine our prediction for budding yeast and find that almost all predicted linkages are supported by further evidence.

Extraction of phylogenetic network modules from the metabolic network.

BackgroundIn bio-systems, genes, proteins and compounds are related to each other, thus forming complex networks. Although each organism has its individual network, some organisms contain common sub-networks based on function. Given a certain sub-network, the distribution of organisms common to it represents the diversity of its function.ResultsWe extracted such "common" sub-networks, defined as "phylogenetic network modules," using phylogenetic profiles and cluster analysis. The enzymes in the same "phylogenetic network module" have similar phylogenetic profiles and related functions. These modules are shown to be phylogenetic building blocks. Furthermore, the network of the modules illustrated hierarchical feature as well as the network of enzymes involved in the metabolism.ConclusionWe conclude that phylogenetic network modules are evolutionary conserved functional units in the metabolic network. We claim that our concept of phylogenetic modules provides a more accurate understanding of the evolution of biological networks.

Assessing the limits of genomic data integration for predicting protein networks

Genomic data integration--the process of statistically combining diverse sources of information from functional genomics experiments to make large-scale predictions--is becoming increasingly prevalent. One might expect that this process should become progressively more powerful with the integration of more evidence. Here, we explore the limits of genomic data integration, assessing the degree to which predictive power increases with the addition of more features. We focus on a predictive context that has been extensively investigated and benchmarked in the past-the prediction of protein-protein interactions in yeast. We start by using a simple Naive Bayes classifier for integrating diverse sources of genomic evidence, ranging from coexpression relationships to similar phylogenetic profiles. We expand the number of features considered for prediction to 16, significantly more than previous studies. Overall, we observe a small, but measurable improvement in prediction performance over previous benchmarks, based on four strong features. This allows us to identify new yeast interactions with high confidence. It also allows us to quantitatively assess the inter-relations amongst different genomic features. It is known that subtle correlations and dependencies between features can confound the strength of interaction predictions. We investigate this issue in detail through calculating mutual information. To our surprise, we find no appreciable statistical dependence between the many possible pairs of features. We further explore feature dependencies by comparing the performance of our simple Naive Bayes classifier with a boosted version of the same classifier, which is fairly resistant to feature dependence. We find that boosting does not improve performance, indicating that, at least for prediction purposes, our genomic features are essentially independent. In summary, by integrating a few (i.e., four) good features, we approach the maximal predictive power of current genomic data integration; moreover, this limitation does not reflect (potentially removable) inter-relationships between the features.

Gene content phylogeny of herpesviruses.

Clusters of orthologous groups [COGs; Tatusov, R. L., Koonin, E. V. & Lipman, D. J. (1997) Science 278, 631-637] were identified for a set of 13 completely sequenced herpesviruses. Each COG represented a family of gene products conserved across several herpes genomes. These families were defined without using an arbitrary threshold criterion based on sequence similarity. The COG technique was modified so that variable stringency in COG construction was possible. High stringencies identify a core set of highly conserved genes. Varying COG stringency reveals differences in the degree of conservation between functional classes of genes. The COG data were used to construct whole-genome phylogenetic trees based on gene content. These trees agree well with trees based on other methods and are robust when tested by bootstrap analysis. The COG data also were used to construct a reciprocal tree that clustered genes with similar phylogenetic profiles. This clustering may give clues to genes with related functions or with related histories of acquisition and loss during herpesvirus evolution.

Phylogenetic analysis of the three polymerase genes (PB1, PB2 and PA) of influenza B virus

Phylogenetic patterns of the three polymerase (PB2, PB1 and PA) genes of a total of 20 influenza B viruses isolated during a 58 year period, 1940-1998, were analysed in detail in a parallel manner. All three polymerase genes consistently showed evolutionary divergence into two major distinct lineages and their amino acid profiles demonstrated conserved lineage-specific substitutions. Dendrogram topologies of the PB2 and PB1 genes were very similar and contrasted with that of the PA gene. It was of particular interest to reveal that even though the PA gene evolved into two major lineages, that of three recent Asian Victoria/1/87-like strains formed a branch cluster located in the same lineage as that of recent Yamagata/16/88-like isolates. Differences in the phylogenetic pathways of three polymerase genes were not only a reflection of genetic reassortment between co-circulating influenza B viruses, but also an indication that the polymerase genes were not co-evolving as a unit. As a result, comparison of the phylogenetic patterns of the three polymerase genes with previously determined patterns of the HA, NP, M and NS genes of 18 viruses defined the existence of eight distinct genome constellations. Also, similar phylogenetic profiles among the PA, NP and M genes, as well as between the PB2 and PB1 genes, were observed, suggesting possible functional interactions among these proteins. Completion of evolutionary analysis of the six internal genes and the HA gene of influenza B viruses revealed frequent genetic reassortment among co-circulating variable strains and suggested co-dependent evolution of genes.