Manikandan Narayanan

Bio: Manikandan Narayanan is an academic researcher from National Institutes of Health. The author has contributed to research in topics: Biology & Gene regulatory network. The author has an hindex of 14, co-authored 23 publications receiving 2054 citations. Previous affiliations of Manikandan Narayanan include Indian Institute of Technology Madras & Bosch.

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

Abstract: With the evolution of multicellularity, communication among cells in different organs/tissues became pivotal to life. Molecular basis of such communication has long been studied, but genome-wide screens for biomolecules/genes 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 excels 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 investigations in humans.

Predicting cross-tissue hormone-gene relations using balanced word embeddings

Abstract: Motivation Large volumes of biomedical literature present an opportunity to build whole-body human models comprising both within-tissue and across-tissue interactions among genes. Current studies have mostly focused on identifying within-tissue or tissue-agnostic associations, with a heavy emphasis on associations among disease, genes and drugs. Literature mining studies that extract relations pertaining to inter-tissue communication, such as between genes and hormones, are solely missing. Results We present here a first study to identify from literature the genes involved in inter-tissue signaling via a hormone in the human body. Our models BioEmbedS and BioEmbedS-TS respectively predict if a hormone-gene pair is associated or not, and whether an associated gene is involved in the hormone’s production or response. Our models are classifiers trained on word embeddings that we had carefully balanced across different strata of the training data such as across production vs. response genes of a hormone (or) well-studied vs. poorly-represented hormones in the literature. Model training and evaluation are enabled by a unified dataset called HGv1 of ground-truth associations between genes and known endocrine hormones that we had compiled. Our models not only recapitulate known gene mediators of tissue-tissue signaling (e.g., at average 70.4% accuracy for BioEmbedS), but also predicts novel genes involved in inter-tissue communication in humans. Furthermore, the species-agnostic nature of our ground-truth HGv1 data and our predictive modeling approach, demonstrated concretely using human data and generalized to mouse, hold much promise for future work on elucidating inter-tissue signaling in other multi-cellular organisms. Availability Proposed HGv1 dataset along with our models’ predictions, and the associated code to reproduce this work are available respectively at https://cross-tissue-signaling.herokuapp.com/, and https://github.com/BIRDSgroup/BioEmbedS. Contact nmanik@cse.iitm.ac.in

Parameterized Algorithms for the Steiner Tree Problem on a Directed Hypercube.

Abstract: We address the problem of computing a Steiner Arborescence on a directed hypercube. The directed hypercube enjoys a special connectivity structure among its node set $\{0,1\}^m$, but its exponential in $m$ size renders traditional Steiner tree algorithms inefficient. Even though the problem was known to be NP-complete, parameterized complexity of the problem was unknown. With applications in evolutionary tree reconstruction algorithms and incremental algorithms for computing a property on multiple input graphs, any algorithm for this problem would open up new ways to study these applications. In this paper, we present the first algorithms, to the best our knowledge, that prove the problem to be fixed parameter tractable (FPT) wrt two natural parameters -- number of input terminals and penalty of the arborescence. Given any directed $m$-dimensional hypercube, rooted at the zero node, and a set of input terminals $R$ that needs to be spanned by the Steiner arborescence, we prove that the problem is FPT wrt the penalty parameter $q$, by providing a randomized algorithm that computes an optimal arborescence $T$ in $O\left(q^44^{q\left(q+1\right)}+q\left|R\right|m^2\right)$ with probability at least $4^{-q}$. If we trade-off exact solution for an additive approximate one, then we can design a parameterized approximation algorithm with better running time - computing an arborescence $T$ with cost at most $OPT+(\left|R\right|-4)(q_{opt}-1)$ in time $O(\left|R\right|m^2+1.2738^{q_{opt}})$. We also present a dynamic programming algorithm that computes an optimal arborescence in $O(3^{\left|R\right|}\left|R\right|m)$ time, thus proving that the problem is FPT on the parameter $\left|R\right|$.

Robust discovery of causal gene networks via measurement error estimation and correction

Abstract: Discovering causal relations among genes from observational data is a fundamental problem in systems biology, especially in humans where direct gene interventions or perturbations are unethical/infeasible. Furthermore, causality is emerging as an integral factor for building interpretable and generalizable machine-learning models of complex phenotypes. Existing methods can discover causal relations from observed gene expression and matched genetic data using the well-established framework of Mendelian Randomization. But, the prevalence of expression measurement errors can mislead most existing methods into making wrong causal discoveries, especially among genes transcribed at low to moderate levels and using data with large sample size (say thousands as in modern genomic or GWAS studies). In this study, we propose a new framework for causal discovery that is robust against measurement noise by extending an established statistical approach CIT (Causal Inference Test). We specifically developed a two-stage approach called RCD (Robust Causal Discovery), wherein we first estimate measurement error from gene expression data and then incorporate it to get consistent parameter estimates that could be used with appropriately extended statistical tests of correlation or mediation done in the original CIT. By quantifying and accounting for noise in the data, our RCD method is able to significantly outperform the baseline method in recovering ground-truth causal relations among simulated noisy genes and transcription factor to target gene relations among noisy yeast genes using data on 1012 yeast segregants. Encouraged by these results, we applied our RCD to a human setting where perturbations are infeasible and identified several causal relations, including ones involving transcriptional regulators in the skeletal muscle tissue. Data and Code Availability The code that implements our two-stage RCD framework is available here: https://github.com/BIRDSgroup/RCD; code for reproducing the figures/tables in this manuscript is also provided in this link.

The Genotype-Tissue Expression (GTEx) pilot analysis: Multitissue gene regulation in humans

Abstract: Understanding the functional consequences of genetic variation, and how it affects complex human disease and quantitative traits, remains a critical challenge for biomedicine. We present an analysi...

Neuroinflammation in Alzheimer's disease

Abstract: Increasing evidence suggests that Alzheimer's disease pathogenesis is not restricted to the neuronal compartment, but includes strong interactions with immunological mechanisms in the brain. Misfolded and aggregated proteins bind to pattern recognition receptors on microglia and astroglia, and trigger an innate immune response characterised by release of inflammatory mediators, which contribute to disease progression and severity. Genome-wide analysis suggests that several genes that increase the risk for sporadic Alzheimer's disease encode factors that regulate glial clearance of misfolded proteins and the inflammatory reaction. External factors, including systemic inflammation and obesity, are likely to interfere with immunological processes of the brain and further promote disease progression. Modulation of risk factors and targeting of these immune mechanisms could lead to future therapeutic or preventive strategies for Alzheimer's disease.

The amyloid hypothesis of Alzheimer's disease at 25 years

Abstract: Despite continuing debate about the amyloid β‐protein (or Aβ hypothesis, new lines of evidence from laboratories and clinics worldwide support the concept that an imbalance between production and clearance of Aβ42 and related Aβ peptides is a very early, often initiating factor in Alzheimer9s disease (AD). Confirmation that presenilin is the catalytic site of γ‐secretase has provided a linchpin: all dominant mutations causing early‐onset AD occur either in the substrate (amyloid precursor protein, APP) or the protease (presenilin) of the reaction that generates Aβ. Duplication of the wild‐type APP gene in Down9s syndrome leads to Aβ deposits in the teens, followed by microgliosis, astrocytosis, and neurofibrillary tangles typical of AD. Apolipoprotein E4, which predisposes to AD in > 40% of cases, has been found to impair Aβ clearance from the brain. Soluble oligomers of Aβ42 isolated from AD patients9 brains can decrease synapse number, inhibit long‐term potentiation, and enhance long‐term synaptic depression in rodent hippocampus, and injecting them into healthy rats impairs memory. The human oligomers also induce hyperphosphorylation of tau at AD‐relevant epitopes and cause neuritic dystrophy in cultured neurons. Crossing human APP with human tau transgenic mice enhances tau‐positive neurotoxicity. In humans, new studies show that low cerebrospinal fluid (CSF) Aβ42 and amyloid‐PET positivity precede other AD manifestations by many years. Most importantly, recent trials of three different Aβ antibodies (solanezumab, crenezumab, and aducanumab) have suggested a slowing of cognitive decline in post hoc analyses of mild AD subjects. Although many factors contribute to AD pathogenesis, Aβ dyshomeostasis has emerged as the most extensively validated and compelling therapeutic target.

Mash: fast genome and metagenome distance estimation using MinHash.

Abstract: Mash extends the MinHash dimensionality-reduction technique to include a pairwise mutation distance and P value significance test, enabling the efficient clustering and search of massive sequence collections. Mash reduces large sequences and sequence sets to small, representative sketches, from which global mutation distances can be rapidly estimated. We demonstrate several use cases, including the clustering of all 54,118 NCBI RefSeq genomes in 33 CPU h; real-time database search using assembled or unassembled Illumina, Pacific Biosciences, and Oxford Nanopore data; and the scalable clustering of hundreds of metagenomic samples by composition. Mash is freely released under a BSD license ( https://github.com/marbl/mash ).