Searcharxiv⌕ Search

arXiv subjects

Balazs Szalkai

Publications and source records attributed to Balazs Szalkai.

12 recordsLinked to original sources
arXiv2017

MetaHMM: A Webserver for Identifying Novel Genes with Specified Functions in Metagenomic Samples

The fast and affordable sequencing of large clinical and environmental metagenomic datasets opens up new horizons in medical and biotechnological applications. It is believed that today we have described only about 1\% of the microorganisms on the Earth, therefore, metagenomic analysis mostly deals with unknown species in the samples. Microbial communities in extreme environments may contain genes with high biotechnological potential, and clinical metagenomes, related to diseases, may uncover still unknown pathogens and pathological mechanisms in known diseases. While the species-level identification and description of the taxa in the samples does not seem to be possible today, we can search for novel genes with known functions in these samples, using numerous techniques, including artificial intelligence tools, like the hidden Markov models (HMMs). Here we describe a simple-to-use webserver, the MetaHMM, which is capable of homology-based automatic model-building for the genes to be searched for, and it also finds the closest matches in the metagenome. The webserver uses already highly successful building blocks: it performs multiple alignment by applying Clustal Omega, builds a hidden Markov model with HMMER components of hmmbuild and uses hmmsearch for finding similar sequences to the specified model in the metagenomes. The webserver is publicly available at \url{https://metahmm.pitgroup.org}.

q-bio.GN↗
arXiv2017

SCARF: A Biomedical Association Rule Finding Webserver

The analysis of enormous datasets with missing data entries is a standard task in biological and medical data processing. Large-scale, multi-institution clinical studies are the typical examples of such datasets. These sets make possible the search for multi-parametric relations since from the plenty of the data one is likely to find a satisfying number of subjects with the required parameter ensembles. Specifically, finding combinatorial biomarkers for some given condition also needs a very large dataset to analyze. For this goal, statistical regression analysis is not the preferred tool of choice, since (i) the {\em a priori} knowledge of the parameter-sets to analyze is missing, and (ii) typically relatively few subjects have the interesting parameter-value ensembles for the analysis. For fast and automatic multi-parametric relation discovery association-rule finding tools are used for more than two decades in the data-mining community. Here we present the SCARF webserver for {\em generalized} association rule mining. Association rules are of the form: $a$ AND $b$ AND ... AND $x \rightarrow y$, meaning that the presence of properties $a$ AND $b$ AND ... AND $x$ implies property $y$; our algorithm finds generalized association rules, since it also finds logical disjunctions (i.e., ORs) at the left-hand side, allowing the discovery of more complex rules in a more compressed form in the database. This feature also helps reducing the typically very large result-tables of such studies, since allowing ORs in the left-hand side of a single rule could include dozens of classical rules. The capabilities of the SCARF algorithm were demonstrated in mining the Alzheimer's database of the Coalition Against Major Diseases (CAMD) in our recent publication (Archives of Gerontology and Geriatrics Vol. 73, pp. 300-307, 2017). Here we describe the webserver implementation of the algorithm.

cs.DB↗
arXiv2017

Comparing Advanced Graph-Theoretical Parameters of the Connectomes of the Lobes of the Human Brain

Deep, classical graph-theoretical parameters, like the size of the minimum vertex cover, the chromatic number, or the eigengap of the adjacency matrix of the graph were studied widely by mathematicians in the last century. Most researchers today study much simpler parameters of braingraphs or connectomes which were defined in the last twenty years for enormous networks -- like the graph of the World Wide Web -- with hundreds of millions of nodes. Since the connectomes, describing the connections of the human brain, typically contain several hundred vertices today, one can compute and analyze the much deeper, harder-to-compute classical graph parameters for these, relatively small graphs of the brain. This deeper approach has proven to be very successful in the comparison of the connectomes of the sexes in our earlier works: we have shown that graph parameters, deeply characterizing the graph connectivity are significantly better in women's connectomes than in men's. In the present contribution we compare numerous graph parameters in the three largest lobes --- frontal, parietal, temporal --- and in both hemispheres of the brain. We apply the diffusion weighted imaging data of 423 subjects of the NIH-funded Human Connectome Project, and present some findings, never described before, including that the right parietal lobe contains significantly more edges, has higher average degree, density, larger minimum vertex cover and Hoffman bound than the left parietal lobe. Similar advantages in the deep graph connectivity properties are hold for the left frontal vs. the right frontal and for the right temporal vs. the left temporal lobes.

q-bio.NC↗
arXiv2017

SECLAF: A Webserver and Deep Neural Network Design Tool for Biological Sequence Classification

Artificial intelligence (AI) tools are gaining more and more ground each year in bioinformatics. Learning algorithms can be taught easily by using the existing enormous biological databases, and the resulting models can be used for the high-quality classification of novel, un-categorized data in numerous areas, including biological sequence analysis. Here we introduce SECLAF, an artificial neural-net based biological sequence classifier framework, which uses the Tensorflow library of Google, Inc. By applying SECLAF for residue-sequences, we have reported (Methods (2017), https://doi.org/10.1016/j.ymeth.2017.06.034) the most accurate multi-label protein classifier to date (UniProt --into 698 classes-- AUC 99.99\%; Gene Ontology --into 983 classes-- AUC 99.45\%). Our framework SECLAF can be applied for other sequence classification tasks, as we describe in the present contribution. Availability and implementation: The program SECLAF is implemented in Python, and is available for download, with example datasets at the website https://pitgroup.org/seclaf/. For Gene Ontology and UniProt based classifications a webserver is also available at the address above.

q-bio.BM↗
arXiv2017

Near Perfect Protein Multi-Label Classification with Deep Neural Networks

Artificial neural networks (ANNs) have gained a well-deserved popularity among machine learning tools upon their recent successful applications in image- and sound processing and classification problems. ANNs have also been applied for predicting the family or function of a protein, knowing its residue sequence. Here we present two new ANNs with multi-label classification ability, showing impressive accuracy when classifying protein sequences into 698 UniProt families (AUC=99.99%) and 983 Gene Ontology classes (AUC=99.45%).

q-bio.BM↗
arXiv2016

The braingraph.org Database of High Resolution Structural Connectomes and the Brain Graph Tools

Based on the data of the NIH-funded Human Connectome Project, we have computed structural connectomes of 426 human subjects in five different resolutions of 83, 129, 234, 463 and 1015 nodes and several edge weights. The graphs are given in anatomically annotated GraphML format that facilitates better further processing and visualization. For 96 subjects, the anatomically classified sub-graphs can also be accessed, formed from the vertices corresponding to distinct lobes or even smaller regions of interests of the brain. For example, one can easily download and study the connectomes, restricted to the frontal lobes or just to the left precuneus of 96 subjects using the data. Partially directed connectomes of 423 subjects are also available for download. We also present a GitHub-deposited set of tools, called the Brain Graph Tools, for several processing tasks of the connectomes on the site \url{http://braingraph.org}.

q-bio.NC↗
arXiv2016

The Dorsal Striatum and the Dynamics of the Consensus Connectomes in the Frontal Lobe of the Human Brain

In the applications of the graph theory it is unusual that one considers numerous, pairwise different graphs on the very same set of vertices. In the case of human braingraphs or connectomes, however, this is the standard situation: the nodes correspond to anatomically identified cerebral regions, and two vertices are connected by an edge if a diffusion MRI-based workflow identifies a fiber of axons, running between the two regions, corresponding to the two vertices. Therefore, if we examine the braingraphs of $n$ subjects, then we have $n$ graphs on the very same, anatomically identified vertex set. It is a natural idea to describe the $k$-frequently appearing edges in these graphs: the edges that are present between the same two vertices in at least $k$ out of the $n$ graphs. Based on the NIH-funded large Human Connectome Project's public data release, we have reported the construction of the Budapest Reference Connectome Server \url{http://connectome.pitgroup.org} that generates and visualizes these $k$-frequently appearing edges. We call the graphs of the $k$-frequently appearing edges "$k$-consensus connectomes" since an edge could be included only if it is present in at least $k$ graphs out of $n$. Considering the whole human brain, we have reported a surprising property of these consensus connectomes earlier. In the present work we are focusing on the frontal lobe of the brain, and we report here a similarly surprising dynamical property of the consensus connectomes when $k$ is gradually changed from $k=n$ to $k=1$: the connections between the nodes of the frontal lobe are seemingly emanating from those nodes that were connected to sub-cortical structures of the dorsal striatum: the caudate nucleus, and the putamen. We hypothesize that this dynamic behavior copies the axonal fiber development of the frontal lobe.

q-bio.NC↗
arXiv2016

Mapping Correlations of Psychological and Connectomical Properties of the Dataset of the Human Connectome Project with the Maximum Spanning Tree Method

We analyzed correlations between more than 700 psychological-, anatomical- and connectome--properties, originated from the Human Connectome Project's (HCP) 500-subject dataset. Apart from numerous natural correlations, which describe parameters computable or approximable from one another, we have discovered numerous significant correlations in the dataset, never described before. We also have found correlations described very recently independently from the HCP-dataset: e.g., between gambling behavior and the number of the connections leaving the insula.

q-bio.NC↗
arXiv2015

Graph Theoretical Analysis Reveals: Women's Brains are Better Connected than Men's

Deep graph-theoretic ideas in the context with the graph of the World Wide Web led to the definition of Google's PageRank and the subsequent rise of the most-popular search engine to date. Brain graphs, or connectomes, are being widely explored today. We believe that non-trivial graph theoretic concepts, similarly as it happened in the case of the World Wide web, will lead to discoveries enlightening the structural and also the functional details of the animal and human brains. When scientists examine large networks of tens or hundreds of millions of vertices, only fast algorithms can be applied because of the size constraints. In the case of diffusion MRI-based structural human brain imaging, the effective vertex number of the connectomes, or brain graphs derived from the data is on the scale of several hundred today. That size facilitates applying strict mathematical graph algorithms even for some hard-to-compute (or NP-hard) quantities like vertex cover or balanced minimum cut. In the present work we have examined brain graphs, computed from the data of the Human Connectome Project, recorded from male and female subjects between ages 22 and 35. Significant differences were found between the male and female structural brain graphs: we show that the average female connectome has more edges, is a better expander graph, has larger minimal bisection width, and has more spanning trees than the average male connectome. Since the average female brain weights less than the brain of males, these properties show that the female brain is more "well-connected" or perhaps, more "efficient" in a sense than the brain of males.

q-bio.NC↗
arXiv2015

The Budapest Reference Connectome Server v2.0

The connectomes of different human brains are pairwise distinct: we cannot talk about an abstract "graph of the brain". Two typical connectomes, however, have quite a few common graph edges that may describe the same connections between the same cortical areas. The Budapest Reference Connectome Server Ver. 2.0 (http://connectome.pitgroup.org) generates the common edges of the connectomes of 96 distinct cortexes, each with 1015 vertices, computed from 96 MRI data sets of the Human Connectome Project. The user may set numerous parameters for the identification and filtering of common edges, and the graphs are downloadable in both csv and GraphML formats; both formats carry the anatomical annotations of the vertices, generated by the Freesurfer program. The resulting consensus graph is also automatically visualized in a 3D rotating brain model on the website. The consensus graphs, generated with various parameter settings, can be used as reference connectomes based on different, independent MRI images, therefore they may serve as reduced-error, low-noise, robust graph representations of the human brain.

q-bio.NC↗
arXiv2013

Identifying Combinatorial Biomarkers by Association Rule Mining in the CAMD Alzheimer's Database

Background: The concept of combinatorial biomarkers was conceived around 2010: it was noticed that simple biomarkers are often inadequate for recognizing and characterizing complex diseases. Methods: Here we present an algorithmic search method for complex biomarkers which may predict or indicate Alzheimer's disease (AD) and other kinds of dementia. We applied data mining techniques that are capable to uncover implication-like logical schemes with detailed quality scoring. Our program SCARF is capable of finding multi-factor relevant association rules automatically. The new SCARF program was applied for the Tucson, Arizona based Critical Path Institute's CAMD database, containing laboratory and cognitive test data for more than 6000 patients from the placebo arm of clinical trials of large pharmaceutical companies, and consequently, the data is much more reliable than numerous other databases for dementia. Results: The results suggest connections between liver enzyme-, B12 vitamin-, sodium- and cholesterol levels and dementia, and also some hematologic parameter-levels and dementia.

q-bio.NC↗