You are considered one of the pioneers of the field of systems biology. How do you believe the discipline has changed or expanded during your career as a professor?

The idea is, just like genome sequencing opened the floodgates of genomics by allowing us to read an entire genome-where you basically press a button and get a string of DNA, which is a text file on your computer of A's, T's, C's, and G's-that operation is now kind of routine in terms of reading and assembling genomes. The key is that the genome is just one tiny part of the entire biological system. DNA is packaged inside a chromosome, which is packed inside of the nucleus, which is inside of cells with all of their machinery like the Golgi apparatus and vacuoles, and of course cells make up different cell types, which make up tissues and organs. Wouldn't it be great if we could press a button and get the structure and function of the entire cell, or the entire functioning organism? That's really when people talk about systems biology, what we are trying to do.

Of course as you start to understand more of the structure and function of the rest of the system outside of the nucleus, then that does shed light on disease, and disease mechanisms. So for COVID-19 and any of these diseases, the idea is that just like DNA sequencing is kind of the core factory operation that one does in lab, it's been automated, there's another set of automated tools we apply called protein interaction mapping, where you identify all protein-protein interactions that interconnect the system. So the genome itself is just a number line, or a string of letters, from which you can find genes; but the genome doesn't necessarily tell you how those genes function together in pathways and protein complexes in cells. And of course almost all genes don't act alone: almost all genes work together to encode parts of molecular machines, kind of like your IKEA manual, where the first page is always the parts list, and pages 2-7 are how you assemble those parts into making the piece of furniture. That's sort of true for any machine that must be assembled; what we work on is what proteins, since proteins are the parts, what proteins are connected to other proteins, and that's what's determined by protein interaction screening. Given all those data, you then want to infer what is that assembled machinery to be revealed.

What is the power of computational/network based strategies in understanding the mechanisms of the novel coronavirus?

So for the COVID example, the idea is you take every viral protein: the viral genome encodes viral genes which make proteins just like our human genome-it's far fewer; most viruses have just a dozen at most proteins, and for those proteins you now perform an affinity purification experiment and you pull proteins down from the human cell, and what you find is that each viral protein is incorporated into multi-protein machines involving both virus and human proteins. That's the sort of tricky thing about viruses, is that they use as much of the human cell as they can. So the papers you read this year on COVID are a map of not just what are the parts of the virus-because that's what the genome tells you-it's a map of how those parts connect to all the parts of human cells to co-opt its function and produce more of the virus. If you're trying to pioneer these approaches which aren't just automatically reading genomes, but are carrying out the rest of the IKEA manual, then viruses are not just important as a public health issue, they're a perfect model system because they're small. And then maybe one day when we figure that out we can scale it up to a whole human cell.

As a co-director of the Cancer Cell Map Initiative [which synergizes the talents of several professors at UCSD and UCSF] , what is your goal for the organization moving forward?

Now that I've explained what we do, we want to do the same thing for cancer. Cancer proteins don't act alone; now that's a very interesting question: is cancer more complicated than viral infection? There certainly are more proteins that have been implicated in cancer than are encoded by a virus. There's hundreds of so-called cancer proteins-what causes a protein to be called a cancer protein? There's lots of reasons, but the most common is that the gene encoding that protein is mutated in most tumors.

So what we're doing is we're starting with each of those mutated proteins in cancer; so far we've looked at 61 genes that are the 'most mutated' in cancer, and what we do is to extract the wiring diagram, so pages 2 through 7 of the furniture assembly manual. We try to develop an exhaustive list of what other proteins those cancer proteins bind to, then use all of this data to really start to see if we can assemble all of the parts of a cancer cell. Some of those parts end up being quite well known-we have a story where collagen is a complex of about 40 different proteins. It turns out that any one of those genes is rarely mutated in tumors, but almost all tumors mutate one of those collagen genes. And you don't see that unless you have the whole machine of collagen, and you're staring at it, and you see that at the machine level, that thing is mutated like gangbusters in most people who have cancer. To understand diseases that are diseases of the genome you need to have that map: so when mutations tinker with that wiring diagram, what are the implications of that? Until you have that wiring diagram, it's going to be hard to make progress.

The 2018 Nobel Prize in Physiology for immunotherapy research sparked incredible interest in pushing forward CAR-T therapies. What do you believe the implications of the Nobel Prize in CRISPR will be on the field of bioengineering?

In the case of CRISPR, I think the impacts have already very much been felt, significantly. It's great that the Nobel Prize Committee recognized both Doudna and Charpentier's contributions earlier-sometimes it will take 40 years when the Committee will realize 'Oh we never gave the Prize for CRISPR'-so it's great that they did that. On the other hand, maybe they don't want to give the prize too soon because they might be worried that CRISPR is just a fad, and it doesn't really work. But I actually agree with the timing of this because it's clear that CRISPR is here to stay and it's utterly changed the field. We (the ldeker Lab) use CRISPR all the time to knock out genes. Once you've assembled the wiring diagram, if proteins A, B, C. and D connect, you need to understand the impact of that machine. CRISPR is a shotgun that basically kicks out different systems and you see what the consequence of having done that is. And to be clear, the ability to snip a gene out of the genome had existed for a while, but it wasn't very efficient. It was much more efficient in model microbes like budding yeast and bacteria; CRISPR makes all of this possible in humans.

Cytoscape (an open source bioinformatics platform] has been widely implemented by systems biologists around the world. In your own words, could you explain how you would like to see scientists incorporate it into their own work?

The success of Cytoscape is really based on not us, but it's everyone out there who uses the tool. The reason why it got heavily used is, first of all, it was timely, so it was the first tool that let you visualize these wiring diagrams that you could read out using protein interaction mapping technologies, along with others. For example, another technology you may have heard about is ChlP-seq (chromatin immunoprecipitation combined with DNA sequencing), which is measuring binding between proteins and DNA as opposed to proteins with other proteins. So it draws a wiring diagram of what transcription factors bind to what gene enhancers and promoters, and Cytoscape can help visualize this.

Second, cytoscape came out around 2003, and over the years we haven't rested; we continually improve the tools and have added lots of new functionality. A lot of software is hard to make work even, especially academic research code that someone publishes with their papers-it's very hard to pick up someone's code and make it useful. But in the case of Cytoscape, we've really invested into making it user friendly and robust, and having a community of users and groups around Cytoscape.

The last thing I'll mention is that we've made it extendible through what we used to call plug-ins but are now called apps. The ability to write an app in Cytoscape is what attracted a lot of developers along with users. The most modern incarnation of that is that rather than having to write your app in a full coding language like Java or C, you write your 'app' in Python or T and use the 'REST communication protocol' (representational state transfer, a method for creating Web services between devices.) to interface to Cytoscpe. You could be running an electronic notebook like Jupyter and writing python code, and have that Python code invade network analysis visualization in Cytoscape.

How do you see cancer bioinformatics research in particular progressing in the future, i.e, new research methodologies or theories/areas of interest?

I think what's going to be possible in the level of automation we've seen with sequencing in DNA and RNA is essentially going to happen with other aspects of biology. So the ability to push a button and just read out the structure of a tumor-now, you can kind of just look at the histopathology of the slide of your tumor cross section and look at cells, but imagine if you could peer into each of those cells and understand its unique structure, and moreover, how that structure differs from the normal cells that surround it. That's going to all become possible, and by extension or as a consequence, treating disease is going to be based on those 'shop' manuals or wiring diagrams of your particular disease. It's like, can you imagine debugging your radio if you didn't have the blueprint?