 Welcome to the dark side of string visualization. In this presentation, I'll cover some of the major problems I see in how people use string networks to visualize omic studies in the literature. If you're not already familiar with the string database, I strongly recommend that you go watch my introduction to the database first. The biggest single problem is what I call zero effort figures. People perform omic studies and find themselves in the situation that they have a long boring list of significantly regulated genes or proteins and they want a colorful figure for their paper. They go to string, query with the list and use the resulting network figure as is in their paper. And while I appreciate the citations that I get this way, if I'm unlucky, they use the evidence view and publish the figure looking something like this. Not only is it aesthetically not pleasing, it doesn't show me anything. If I'm a little bit lucky, they might have used the confidence view instead and maybe even disabled the miniature structure images. It gives you a figure looking like this instead, which I hope you agree is aesthetically a little bit more pleasing. The problem is it still shows absolutely nothing and specifically it doesn't show the data of the omic study. The next step up is pointless figures. These differ from zero effort figures in that some effort has been made. Typically people use cytoscape to import a string network, improve the network layout over the default, import omics data and visualize these data on the network, giving them a figure looking more like this. This is definitely better. It's improved in several ways. Aesthetically it looks better, it is not as colorful anymore, it doesn't have overlapping nodes and more importantly it shows the actual data from the study. The problem is that it's still a hairball. The authors cannot see any patterns in the data from this. They do nothing with the network, they draw no conclusions from the network analysis and the figure ultimately serves no purpose whatsoever other than being there and being colorful. So what can we do about this? The first thing is to be aware of the different network types in string. Do you want a functional association network or do you want perhaps a physical interaction network instead? It's important to make a conscious choice. If you're working with a list of genes where you already know that they all work together, a functional association network will be a hairball and bring nothing new to the table. So maybe use a physical interaction network instead. You can also use a higher confidence cut-off especially if you're working with genes or processes or organisms that are highly studied. That way you will get fewer edges leading to generally better network visualizations. You can use network clustering to cut up the dentist networks and thereby identify functional modules or physical protein complexes depending on the type of network you started from. You can then interpret these, annotate them in terms of what they are and make perhaps separate visualizations of selected subnetworks of interest and most importantly make the figures so that they help you tell a story in the paper. If you want to learn how to do all of this, we have a lot of free training material available to you. I've already made presentations covering the general topic of network visualization, how to use the Cytoscape platform and specifically how to use Cytoscape string app to visualize string networks in Cytoscape. Also we have hands-on exercises available on this web page. Thanks for your attention. This is all I want to say about this topic this time. If you want to hear me complain about enrichment analysis as well, I recommend that you go watch this presentation next. Thanks for your attention.