Alzheimer's disease rewires gene coexpression networks coupling different brain regions.

Connectome studies have shown how Alzheimer's disease (AD) disrupts functional and structural connectivity among brain regions. But the molecular basis of such disruptions is less studied, with most genomic/transcriptomic studies performing within-brain-region analyses. To inspect how AD rewires the correlation structure among genes in different brain regions, we performed an Inter-brain-region Differential Correlation (Inter-DC) analysis of RNA-seq data from Mount Sinai Brain Bank on four brain regions (frontal pole, superior temporal gyrus, parahippocampal gyrus and inferior frontal gyrus, comprising 264 AD and 372 control human post-mortem samples). An Inter-DC network was assembled from all pairs of genes across two brain regions that gained (or lost) correlation strength in the AD group relative to controls at FDR 1%. The differentially correlated (DC) genes in this network complemented known differentially expressed genes in AD, and likely reflects cell-intrinsic changes since we adjusted for cell compositional effects. Each brain region used a distinctive set of DC genes when coupling with other regions, with parahippocampal gyrus showing the most rewiring, consistent with its known vulnerability to AD. The Inter-DC network revealed master dysregulation hubs in AD (at genes ZKSCAN1, SLC5A3, RCC1, IL17RB, PLK4, etc.), inter-region gene modules enriched for known AD pathways (synaptic signaling, endocytosis, etc.), and candidate signaling molecules that could mediate region-region communication. The Inter-DC network generated in this study is a valuable resource of gene pairs, pathways and signaling molecules whose inter-brain-region functional coupling is disrupted in AD, thereby offering a new perspective of AD etiology.

MultiCens: Multilayer network centrality measures to uncover molecular mediators of tissue-tissue communication.

With the evolution of multicellularity, communication among cells in different tissues and organs became pivotal to life. Molecular basis of such communication has long been studied, but genome-wide screens for genes and other biomolecules mediating tissue-tissue signaling are lacking. To systematically identify inter-tissue mediators, we present a novel computational approach MultiCens (Multilayer/Multi-tissue network Centrality measures). Unlike single-layer network methods, MultiCens can distinguish within- vs. across-layer connectivity to quantify the "influence" of any gene in a tissue on a query set of genes of interest in another tissue. MultiCens enjoys theoretical guarantees on convergence and decomposability, and performs well on synthetic benchmarks. On human multi-tissue datasets, MultiCens predicts known and novel genes linked to hormones. MultiCens further reveals shifts in gene network architecture among four brain regions in Alzheimer's disease. MultiCens-prioritized hypotheses from these two diverse applications, and potential future ones like "Multi-tissue-expanded Gene Ontology" analysis, can enable whole-body yet molecular-level systems investigations in humans.

Decoding the 5' nucleotide bias of PIWI-interacting RNAs.

PIWI-interacting RNAs (piRNAs) are at the center of a small RNA-based immune system that defends genomes against the deleterious action of mobile genetic elements (transposons). PiRNAs are highly variable in sequence with extensive targeting potential. Their diversity is restricted by their preference to start with a Uridine (U) at the 5' most position (1U-bias), a bias that remains poorly understood. Here we uncover that the 1U-bias of Piwi-piRNAs is established by consecutive discrimination against all nucleotides but U, first during piRNA biogenesis and then upon interaction with Piwi's specificity loop. Sequence preferences during piRNA processing also restrict U across the piRNA body with the potential to directly impact target recognition. Overall, the uncovered signatures could modulate specificity and efficacy of piRNA-mediated transposon restriction, and provide a substrate for purifying selection in the ongoing arms race between genomes and their mobile parasites.

Decrypting ENCODEd epigenetic marks of human tRN-A-RS genes in normal, stem and cancer cell lines.

Screening large-scale ENCODE data of 625 cytoplasmic transfer RNA (tRNAs) and 37 aminoacyl tRNA synthetase (AARSs) human genes, we deconstruct the array of relations between 10 histone marks affecting 15 chromatin states; their tissue specificity and variations and interchange amongst normal, cancerous and stem cells. The histone marks of RNA Pol II transcribed AARS genes share, but also contrast with that on RNA Pol III transcribed tRNA genes. tRNAs with identical/similar sequences may be in significantly varying states even within the same cell line; the chromatin scaffold, where the tRNA gene resides, is the key determinant. Hepatocellular carcinoma cell line has dominant H3K27me3, and singular clustering of other marks. Leukaemic cell line has hyperactive genes. The quiescence of the stem cells is encoded in the markers. Leaving aside the important exceptions in stem cells and elsewhere, tRNAs with cove scores above 50 have active markers and precise sets of transcription factors, and are usually well conserved compared to the low-scoring ones. Pseudo tRNAs are in heterochromatin/repressed state with anomalous exceptions in cancer cells. We motivate that Epigenetic-Phishing hacks the translation apparatus through the chromatin states governed by the histone marks of tRNA and AARS genes, and speculate on their therapeutic implications in cancer and on stem cells.

Eukaryotic tRNA paradox.

tRNAs are widely believed to segregate into two classes, I and II. Computational analysis of eukaryotic tRNA entries in Genomic tRNA Database, however, leads to new, albeit paradoxical, presence of more than a thousand class-I tRNAs with uncharacteristic long variable arms (V-arms), like in class-II. Out of 62,202 tRNAs from 69 eukaryotes, as many as 1431 class-I tRNAs have these novel extended V-arms, and we refer to them as paradoxical tRNAs (pxtRNAs). A great majority of these 1431 pxtRNA genes are located in intergenic regions, about 18% embedded in introns of genes or ESTs, and just one in 3'UTR. A check on the conservations of 2D and 3D base pairs for each position of these pxtRNAs reveals a few variations, but they seem to have almost all the known features (already known identity and conserved elements of tRNA). Analyses of the A-Box and B-Box of these pxtRNA genes in eukaryotes display salient deviations from the previously annotated conserved features of the standard promoters, whereas the transcription termination signals are just canonical and non-canonical runs of thymidine, similar to the ones in standard tRNA genes. There is just one such pxtRNA(ProAGG) gene in the entire human genome, and the availability of data allows epigenetic analysis of this human pxtRNA(ProAGG) in three different cell lines, H1 hESC, K562, and NHEK, to assess the level of its expression. Histone acetylation and methylation of this lone pxtRNA(ProAGG) gene in human differ from that of the nine standard human tRNA(ProAGG) genes. The V-arm nucleotide sequences and their secondary structures in pxtRNA differ from that of class-II tRNA. Considering these differences, hypotheses of alternative splicing, non-canonical intron and gene transfer are examined to partially improve the Cove scores of these pxtRNAs and to critically question their antecedence and novelty.

Eukaryotic tRNAs fingerprint invertebrates vis-à-vis vertebrates.

During translation, aminoacyl-tRNA synthetases recognize the identities of the tRNAs to charge them with their respective amino acids. The conserved identities of 58,244 eukaryotic tRNAs of 24 invertebrates and 45 vertebrates in genomic tRNA database were analyzed and their novel features extracted. The internal promoter sequences, namely, A-Box and B-Box, were investigated and evidence gathered that the intervention of optional nucleotides at 17a and 17b correlated with the optimal length of the A-Box. The presence of canonical transcription terminator sequences at the immediate vicinity of tRNA genes was ventured. Even though non-canonical introns had been reported in red alga, green alga, and nucleomorph so far, fairly motivating evidence of their existence emerged in tRNA genes of other eukaryotes. Non-canonical introns were seen to interfere with the internal promoters in two cases, questioning their transcription fidelity. In a first of its kind, phylogenetic constructs based on tRNA molecules delineated and built the trees of the vast and diverse invertebrates and vertebrates. Finally, two tRNA models representing the invertebrates and the vertebrates were drawn, by isolating the dominant consensus in the positional fluctuations of nucleotide compositions.

Anomalous altered expressions of downstream gene-targets in TP53-miRNA pathways in head and neck cancer.

The prevalence of head and neck squamous cell carcinoma, HNSCC, continues to grow. Change in the expression of TP53 in HNSCC affects its downstream miRNAs and their gene targets, anomalously altering the expressions of the five genes, MEIS1, AGTR1, DTL, TYMS and BAK1. These expression alterations follow the repression of TP53 that upregulates miRNA-107, miRNA- 215, miRNA-34 b/c and miRNA-125b, but downregulates miRNA-155. The above five so far unreported genes are the targets of these miRNAs. Meta-analyses of microarray and RNA-Seq data followed by qRT-PCR validation unravel these new ones in HNSCC. The regulatory roles of TP53 on miRNA-155 and miRNA-125b differentiate the expressions of AGTR1 and BAK1in HNSCC vis-à-vis other carcinogenesis. Expression changes alter cell cycle regulation, angiogenic and blood cell formation, and apoptotic modes in affliction. Pathway analyses establish the resulting systems-level functional and mechanistic insights into the etiology of HNSCC.

Viral/plasmid captures in Crenarchaea.

tRNA genes are the integration sites of viral/plasmid genomes into their hosts chromosomes by homologous recombination catalyzed by integrases. The crossover between viral/plasmid and host genomes leaves 3'-fractional tRNA motif as tell-tale marker of integration on host-chromosome. This 3'-fractional tRNA motif on host genome is our retrenched tRNA (rtRNA). To track integration in Crenarchaea, host rtRNAs, and conserved features in viral/plasmid tRNA motifs and in integrases were identified. The viral-integrase has a conserved 24-nucleotide long motif, GTATTATGTTTACTCAATAGAGAA in the N-terminal region. Upstream of the viral tRNA motif has a conserved poly-cytosine region and a hairpin secondary structure. Corresponding to a host tRNA, we observe up to two rtRNAs on crenarchaeal chromosome. The length of the rtRNA is not random. The fraction of tRNA excised off in rtRNA is either 61.8, or 50, or 38.2, or 23.6%. Thus, the integration fragments the tRNA nonrandomly dividing it approximately in ratios 3:2, or 1:1, or 2:3, or 1:3. More than 79% of rtRNAs have lengths that are excised 38.2% off tRNA. It turns out that 38.2% excision implies that the ratio of the length of tRNA to its rtRNA is just 1.618, the golden ratio. Hence, the vast majority of rtRNAs are at or near the golden ratio. Evidence emerges of new extremophile viral entities.

Systems biology of cancer biomarker detection.

Cancer systems-biology is an ever-growing area of research due to explosion of data; how to mine these data and extract useful information is the problem. To have an insight on carcinogenesis one need to systematically mine several resources, such as databases, microarray and next-generation sequences. This review encompasses management and analysis of cancer data, databases construction and data deposition, whole transcriptome and genome comparison, analysing results from high throughput experiments to uncover cellular pathways and molecular interactions, and the design of effective algorithms to identify potential biomarkers. Recent technical advances such as ChIP-on-chip, ChIP-seq and RNA-seq can be applied to get epigenetic information transformed into a high-throughput endeavour to which systems biology and bioinformatics are making significant inroads. The data from ENCODE and GENCODE projects available through UCSC genome browser can be considered as benchmark for comparison and meta-analysis. A pipeline for integrating next generation sequencing data, microarray data, and putting them together with the existing database is discussed. The understanding of cancer genomics is changing the way we approach cancer diagnosis and treatment. To give a better understanding of utilizing available resources' we have chosen oral cancer to show how and what kind of analysis can be done. This review is a computational genomic primer that provides a bird's eye view of computational and bioinformatics' tools currently available to perform integrated genomic and system biology analyses of several carcinoma.

HNOCDB: a comprehensive database of genes and miRNAs relevant to head and neck and oral cancer.

In spite of the wide prevalence of head, neck and oral cancer, HNOC, there is no integrated database on genes and miRNAs associated with all the carcinoma subtypes of HNOC. The objective is to compile a multilayered and comprehensive database of HNOC as a user-friendly resource for researchers devising novel therapeutic strategies. We present HNOCDB, the head, neck and oral cancer database, with the following key features: (i) it tabulates all the different categories of HNOC separately under appropriate subtype-names, and then puts them together in a table headlined All; (ii) the oncogenes/oncomiRs that cause HNOC are listed; their mutations, methylations and polymorphisms loci are marked, and the variations in their expression profiles relative to the normal are recorded; (iii) HNOCDB contains a chromosomal map of HNOC genes and miRNA; (iv) contains references that experimentally validate the reason for the inclusion of the genes and the miRNAs in HNOCDB. HNOCDB is freely accessible for academic and non-profit users via http://gyanxet.com/hno.html.

Antagomirbase- a putative antagomir database.

UNLABELLED: The accurate prediction of a comprehensive set of messenger putative antagomirs against microRNAs (miRNAs) remains an open problem. In particular, a set of putative antagomirs against human miRNA is predicted in this current version of database. We have developed Antagomir database, based on putative antagomirs-miRNA heterodimers. In this work, the human miRNA dataset was used as template to design putative antagomirs, using GC content and secondary structures as parameters. The algorithm used predicted the free energy of unbound antagomirs. Although in its infancy the development of antagomirs, that can target cell specific genes or families of genes, may pave the way forward for the generation of a new class of therapeutics, to treat complex inflammatory diseases. Future versions need to incorporate further sequences from other mammalian homologues for designing of antagomirs for aid in research. AVAILABILITY: The database is available for free at http://bioinfopresidencycollegekolkata.edu.in/antagomirs.html.

Novel hybrid encodes both continuous and split tRNA genes?

tRNAs are mostly transcribed from un-fragmented genes, but occasionally also from split genes, with separated 5' and 3' halves. A reanalysis of the existing data on Staphylothermus marinus and Staphylothermus hellenicus hints of a novel hybrid gene that encodes both an un-fragmented and a 5'-split-half together in one. The corresponding 3' complement-gene is located elsewhere on the genome. As un-fragmented, the hybrid gene transcribes to tRNA(lys)(TTT). But as 5'-half, it trans-splices with its 3'-complement-half to tRNA(lys)(CTT), the tRNA missed so far. This hybrid of the split and the un-fragmented in one suggests a deeper synergy between the two, and hints of co-evolution. Furthermore, in a subtle contrast to the widely held idea of conservation of 3'-half, it is precisely the 3'-half that varies in these two tRNAs; the 5'-half remains conserved.

Multiply expressed tRNA genes?

Some coding (mRNA) genes are known to express into several different amino-acid chains by intron and exon shuffling. There are computational arguments that an analogous mechanism may give rise to two different products from a single non-coding gene. We propose, based on bioinformatics' evidence, that the number of different products from some of these non-coding genes may indeed be greater than two. The present study demonstrates that, in some cases, intron repositioning and partial exon-intron shuffling lead to newer putative tRNAs. An intricate and entangled organizational network performs a complex optimization of the secondary structures at the exon-intron boundaries. We find up to four different RNAs are encoded cryptically in a single composite tRNA gene. But it is the remarkably high fidelity of the secondary structures and the conserved sequences of all the tRNAs that are embedded that leads to this hypothesis.