1<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"> 2<html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en"> 3<head> 4 <meta http-equiv="content-type" content="text/html; charset=utf-8" /> 5 <meta name="description" content="Your description goes here" /> 6 <meta name="keywords" content="your,keywords,goes,here" /> 7 <meta name="author" content="Your Name" /> 8 <link rel="stylesheet" type="text/css" href="variant-trio.css" title="Variant Trio" media="screen,projection" /> 9 <title>Research Projects</title> 10</head> 11 12<body> 13<div id="wrap"> 14<h1><a href="index.html">Schulz Lab</a></h1> 15 <div id="slogan"><h3> Projects</h3> </div> 16 <div id="topmenu"> 17 <p> 18 <strong class="hide">Main menu:</strong> 19 <strong class="hide">Main menu:</strong> 20 <a href="index.html">Home</a><span class="hide"> | </span> 21 <a href="team.html">Team</a><span class="hide"> | </span> 22 <a class="active" href="projects.html">Projects</a><span class="hide"> | </span> 23 <!--<a href="projects.html">Software</a><span class="hide"> | </span> --> 24 <a href="publications.html">Publications</a><span class="hide"> | </span> 25 <a href="teaching.html">Teach</a><span class="hide"> | </span> 26 <a href="software.html">Tools</a><span class="hide"> | </span> 27 <a href="jobs.html">Jobs</a><span class="hide"> | </span> 28 29 30 </p> 31 </div> 32 33 <div id="bigcontent"> 34 35 36 <div class="fullCenter"> 37 38 39 40 <h3>| 41 <a href="#Precision">AI Precision Therapy</a> | 42 <a href="#integrative">Integrative Gene Regulation</a> | 43 <a href="#Triplex">Learning RNA interactions</a> | 44 </h3> 45 <br> 46 47</div> 48 49<div class="fullCenter"> 50 <a name="Precision"></a><br> 51 <h2>AI-based precision therapy for heart failure patients </h2> 52 53 <p align="center"><img src="img/Precision.png" width="650px" alt="Integrative Machine Learning approaches for gene regulation" 54 title="Integrative Machine Learning approaches for gene regulation"/> 55 </p> 56 57<p align="justify"> 58 59Recent studies show that individual hematopoietic stem/progenitor cells (HSPCs) accumulate somatic mutations as a function of age in healthy individuals. Once a mutation has reached a clone size larger than 2% of the variant allele frequency measured with genomic sequencing, it is called clonal hematopoiesis of indeterminate potential (CHIP). CHIP has been associated to chronic heart failure and atherosclerosis. 60Investigating a local cohort of chronic heart failure (CHF) patients, in cooperation with the cardiology department, we have studied the co-occurrence of inherited single nucleotide polymorphisms (SNPs) with the occurrence of CHIP in the genes DNMT3A and TET2. We found a number of inherited SNPs that are associated with a change in survival in the context of CHF patients with CHIP. We are using multi-modal autoencoders to create a joint latent space for the integration of drug response data with single cell RNA-seq data to determine FDA-approved drugs that could provide novel therapies for heart failure treatment. We warmly acknowledge the funding of the computational work on the project by: 61<img src="img/Kassel.jpg" width="290px" alt="Alfons und Gertrud Kassel Stiftung Logo" 62 title="Alfons und Gertrud Kassel Stiftung"/> 63</p></div> 64 65 66 67<div class="fullCenter"> 68 <a name="integrative"></a><h2>Integrative Methods for the Analysis of Gene Regulation</h2> 69 70 <p align="center"><img src="img/Integrative.jpg" width="550px" alt="Integrative Machine Learning approaches for gene regulation" 71 title="Integrative Machine Learning approaches for gene regulation"/> 72 </p> 73 74<p align="justify"> 75Despite the vast amount of research in that area, the regulation of genes is still not fully understood. 76In particular the interplay between regulators that modulate the transcriptional activation of a gene and 77post-transcriptional regulators that modulate the abundance of a geneâs product, the mRNA, is neglected in 78most system biology studies. With the availability of numerous complete epigenomics datasets our vision is 79to produce a comprehensive computational catalogue of gene regulation for each gene in the human genome, 80including transcriptional and post-transcriptional regulators, at much higher detail as is currently available. 81We greatly acknowledge funding from : 82<img src="img/dzhk.png" width="190px" alt="DZHK Logo" 83 title="Deutsches Zentrum für Herz-Kreislauf-Forschung"/> <img src="img/DFG.png" width="190px" alt="DFG Logo" 84 title="Deutsche Forschungsgemeinschaft"/> 85</p> 86<br> 87 88<h3>Prioritization of disease relevant transcriptional regulators from single cell data using interpretable machine learning</h2> 89 90<img src="img/GAZE.png" width="500px" alt="GAZE image" title="GAZE overview"/> 91<p align="justify"> 92Recent technical developments have enabled the profiling of an individualâs cell transcriptome and epigenome. While single cell measurements provide high resolution and allow the study of gene expression and epigenome changes, they provide noisy and sparse measurements of a cellâs activity, in particular for the commonly used 10X sequencing platform. We have been developing machine learning methods for the analysis of single cell RNA and epigenome data that allow prioritization of transcriptional regulators of interest. 93We have developed a statistical framework with an R shiny-based application, called GAZE, that guarantees a comprehensive analysis of single-cell data in an integrative fashion. Using dimensionality reduction and unsupervised machine learning, sparsity of single cell data is reduced by aggregation to so-called metacells. Then supervised sparse regression is used to learn associations between TF binding predictions in enhancer/promoters and gene expression in a per-gene fashion extending our tree-guided multitasking method to associate transcription factor regulation of genes using single cell data. 94 95</p> 96 97 98<h3>Reconstruction of dynamic regulatory networks </h3> 99<p align="justify"> 100Current models of gene regulatory networks are often constructed as a static snapshot of the regulatory wiring 101in cells. We are working on methods that can dynamically rewire the network connections modeling transcriptional 102and posttranscriptional factors through the integration of binding data (e.g. Chip-Seq) and gene expression data. 103In addition, we are enhancing these methods to utilize 104transcript expression level measurements with RNA-Seq to improve the resolution for reconstruction of dynamic regulatory networks. 105</p> 106<p align="center"><img src="img/overview.png" style="right" width="500px" alt="dynamic networks" title="dynamic networks"/> 107 108<p > 109D Gérard, <b>F Schmidt</b>, A Ginolhac, M Schmitz, R Halder, P Ebert, <b>
109MH Schulz</b>, T Sauter, L Sinkkonen 110<br><b><FONT COLOR=#333333>Temporal epigenomic profiling identifies AHR and GLIS1 as super-enhancer controlled regulators of mesenchymal multipotency</b>,<br> 111<i> Nucleic Acids Research 2019 <a href=https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gky1240/5245443> [full text]</a> 112</p> 113<p ><b> MH Schulz</b>, KV Pandit, CL Lino Cardenas, N Ambalavanan, N Kaminski and Z Bar-Joseph <br> 114<b><FONT COLOR=#333333>Reconstructing dynamic microRNA-regulated interaction networks</FONT></b> 115<br> 116<i>PNAS 2013</i> [<a href=http://www.pnas.org/content/early/2013/08/28/1303236110.abstract>full text</a>]<br> 117 118</div> 119 120 121<div class="fullCenter"> 122 <a name="Triplex"></a> 123<h2>An unbiased approach reveals a universal code for RNA·DNA:DNA interaction</h2> 124<p align="justify"> 125 <FONT COLOR=#000000> 126RNA·DNA:DNA triple helix (triplex) formation is a form of RNA-DNA interaction, which regulates gene expression, but is difficult to study experimentally in vivo. This makes accurate computational prediction of such interactions highly important in the field of RNA research. Current predictive methods use canonical Hoogsteen base pairing rules, which whilst biophysically valid, may not reflect the plastic nature of cell biology. We have developed the first unbiased optimization approach to learn a probabilistic model describing RNA-DNA interactions directly from motifs derived from triplex sequencing data. We find that there are several stable interaction codes, including Hoogsteen base pairing and novel RNA-DNA base pairings, which agree with in vitro measurements of triplex binding. We implemented these findings in TriplexAligner, a program that uses the determined interaction codes to predict triplex binding. TriplexAligner predicts RNA-DNA interactions identified in all-to-all sequencing (RADICL-seq) data more accurately than all previously published tools, and also predicts previously studied triplex interactions with known regulatory functions. Our work is an important step towards better understanding of triplex formation and allows genome-wide analyses of RNA-DNA interactions. <br> 127 We acknowledge funding by <img src="img/TRR267.jpeg" style="right" width="190px" alt="TRR267 logo" title="TRR267"/> 128</p> 129<p align="center"><img src="img/Triplexaligner.png" style="right" width="700px" alt="dynamic networks" title="dynamic networks"/> 130<p > 131T Warwick, et al. 132<br><b><FONT COLOR=#333333> 133A universal model of RNA.DNA:DNA triplex formation accurately predicts genome-wide RNA-DNA interactions</b>,<br></FONT> 134<i> Briefings in Bioinformatics, 2022 <a href=https://academic.oup.com/bib/article/23/6/bbac445/6760135?login=false> [full text]</a> 135</p> 136</div> 137 138 139 <hr class="clear" /> 140 </div> 141 142 <div id="footer"> 143 <p>Copyright © 2023 <a href="index.html">Marcel H. Schulz</a><br /> 144 Template design by <a href="http://andreasviklund.com/">Andreas Viklund</a></p> 145 </div> 146 147</div> 148</body> 149</html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.