 String frequently asked question. What is the best confidence card off for my network? This is by far the most common question they get from users and the answer is there is no best card off It depends on what you want to do, but I can give you some rules of thumb When doing network visualization of some hundreds of genes Producing a human functional network I will typically use a confidence card off in the range of 0.8 to 0.9 and then do network clustering to simplify the visualization I will use a lower card off for a physical network because the network is sparser and I will use a lower card off for other Organisms because they're less studied. However in the end choose whatever looks good for your network When doing network analysis my recommendation is very different I recommend using no card off, but using a probabilistic framework that correctly accounts for the confidence scores in string Not treating all edges as equal. That way you make use of all available data in your analysis