Hi everyone,

Thanks for making it this far — welcome to the last Capsid of the year!

I don’t know about you all, but this year has hit a bit differently for us, for better or worse. We’ve been trying new things, been trying to find our footing, been trying to settle in, but instead of settling in, the year has remained quite un-settling.

There’s a lot we looked forward to this year, from the city, from ourselves, from the unknown, and I think looked a bit too forward and fell off the deep end. Maybe our expectations were a bit too high?

We’ve been doing — trying — a lot of things. We’ve definitely learned a ton. If we’d merely expected us to learn new things then by all means, we’ve blown past that metric. Paradoxically, I think we tried more things this year than we’ve previously tried, but at the same time, I feel like we haven’t gotten as much done as we’ve hoped. There’s a lot of experiments we tried but scrapped along the way. But maybe that’s what comes with the territory. If your experiments all go right, you’re not experimenting hard enough.

With that said, there are still plenty of things we’ve done this year, that we’re proud of!

~ Jan

Our favorite Capsids

Ok to start off, one of my favorite Capsids posts is by far the March Phage Picks. This is kind of a cheat, because the post itself isn’t actually my favorite Phage Picks, and those papers aren’t actually my absolute favorite papers. I just liked that we tried something fresh — and according to our email analytics (very few have actually written to use about liking this new format…), readers seemed to like it!

Our Phage Picks for March 2024!

This was a format I’ve wanted to do for quite some time. Both Jess and I have a huge log of favorite papers, and we’ve always wondered how to share our reading stack. About once a month, we share our favorite papers. Mine skew bioinformatics and machine learning, while Jess skews more phage biology. We have lots of ideas of how to improve on the format next year (plus figuring out ways to encourage our readers to share more of their favorite papers, as that’s been the hardest part…).

Building a SEED library for One Health

My other favorite article was this one from Jonathan Moore. It’s just so out there and not like anything else I’ve read in phages, but if you read the article a lot of it is grounded in lots of our existing understanding of biology, and it’s not actually that science fiction. I like this article because it reminds us that there’s a whole lot more applications out there for phages other than therapy.

~ Jan

Phage therapy is unfettered by borders

I loved this article and what it represents. This is an account from a Hungarian PhD student who helped screen and find phages for a patient we were trying to treat in Australia before we left.

~ Jess

PhageBase: A visual atlas of phages

This was a super cool initiative from an undergrad student I worked with in the Bollyky lab this year; he created this all on his own, and has a small team of fellow students manually curating TEM images from phage papers.

~ Jess

Our favorite Papers

I’ve been spending a lot of time reading up on and trying to understand machine learning this year. I’ve also been trying to read up on the space between machine learning and phage biology of which there are a surprisingly high number of both interesting and accessible papers.

In Phage Genome Annotation: Where to Begin and End, Anastasiya and Andy Millard writes the best paper to figure out where phage bioinformatics is today, and where it’s headed.

If you’re new (or catching up) with phage bioinformatics, you absolutely need to start here. Few papers blow my mind anymore, but this paper Deep model predictive control of gene expression in thousands of single cells changed how I thought about ML’s relationship with biology.

And if you’re wanting to figure out phages and machine learning, you absolutely need to read Dimi’s paper Prediction of Klebsiella phage-host specificity at the strain level, where it’s not just a paper but basically an instruction manual on how to set it up. It even includes very well-documented code, and all the data, for you to replicate over a weekend. He makes every piece so incredibly easy to understand.

~ Jan

I too have been trying to understand machine learning this year. Hard not to when you’re living in the physical global hub of AI, and every other week there is a meetup where you meet people working on applying it to biology. And when each day you’re using better and better AI tools for everything else… it’s hard not to start thinking it could be mega powerful if applied to actual scientific problems. (FYI if you’re not using Claude for reading/summarizing/thinking/writing/editing in science, and still relying on ChatGPT, you are missing out).

I personally got really inspired by the Evo paper by Eric Nguyen and colleagues at the Arc Institute (next door to us at Stanford!). The paper used phage data, without labels (ie annotations), to train a model that could predict new made-up protein-RNA complexes, like CRISPR and transposons, that actually work in the lab, and to even predict new phage genomes (that don’t exist, but should be functional, based on what it learned about the other genomes). That lab is now trying to synthesize these genomes to see if their model really does predict the rules for a phage genome. They are starting with phiX174, which is a small and simple one (the first phage/organism sequenced, I believe?).

This led me to Dimi’s paper (the one Jan mentioned above) on predicting strain-level phage-host interactions for Klebsiella, and then the one by Aude Bernheim’s group, led by Baptiste Gaborieau, doing something similar for E. coli. I already wrote about these in last week’s Phage Picks, but needless to say, they inspired me to start really digging into all this stuff. I even made phage + AI (and reorienting our research around using these new tools) into a focus of my recent talk at Phage Option in Colombia (which was an AMAZING time by the way; think the Evergreen of South America…. such lovely, lovely people and a relaxed, sunny vibe!). Anyway, much more on that soon; for now I need to spend Christmas break watching Coursera/Kaggle courses on machine learning 101 so I can figure out HOW to make it useful for phage projects. Stay tuned… and let me know if you’d be interested in joining a ‘Phage ML’ club where we help each other learn this stuff and start designing experiments/projects that generate data that can be fed into ML models. I really think this is what’s next for the phage field.

~ Jess

Jess’s year in reflection

This year I worked with Paul Bollyky’s lab to try to get a center for phage therapy and phage delivery off the ground (both getting it going physically, and getting the funding for it). His lab is interested in using phages for patients but also for delivering therapeutic genes to eukaryotic cells. I was super excited to join, because I wanted to a) see if I could set up a phage therapy center a second time, to prove the model of ‘academic lab + a set of protocols = patients treated’ is repeatable, b) learn how to write NIH grants from someone with a history of getting them, and c) work with really smart people who would push me to think bigger and take my work to the next level.

Walking up to the iconic Stanford campus…

Well, it’s been a pretty crazy year. I haven’t accomplished everything I wanted, but I’ve learned a lot, as Jan mentioned above. Here’s a few thoughts on the year:

Lessons in unseen advantages and unwritten requirements

I definitely learned a lot this year about the unwritten requirements for phage production for patients. Coming into this year I thought it was mostly about knowing what to do, and having the freedom to do it. I felt confident I had both of those things coming into this new role at Stanford — and I do! But I’ve found there are many additional ingredients to a ‘phage therapy center’ (my working definition: a lab that repeatably makes phages for patients) that I didn’t account for. Things I happened to have access to in Australia that has taken me longer than I anticipated to find/establish here (and would be the case anywhere, I think). I’m working on making this all a reality here, but here’s a few things I think should be added to the list of ‘phage therapy center’ ingredients:

Full time personnel dedicated to phage therapy lab work: In Australia I wasn’t writing grants or papers, wasn’t mentoring students. I was full time dedicated to phage therapy production. I was also teamed up with Stephanie Lynch, a phage PhD dedicated to the lab-side of setting up our phage therapy system, who was willing to spend time around the clock in the lab. Not only did she create at least half of the systems, she was there to bounce ideas off regarding labeling, banking phages, workflows, etc. This gave me so much time (and energy) to think about what to do, document what we were doing, and write about it, and the work didn’t grind to a halt if I wasn’t personally able to be in the lab much on a given day (or if my 1.5-hour train commute turned into a 3-hour one a few times every couple weeks due to spontaneous delays). In short, a phage therapy center needs a Stephanie.

Lab space & equipment dedicated to phage therapy work: Lab space is always hard to come by, and that is especially true here (you are actually not allowed to build new buildings on this campus without knocking one down, I have heard?). And there are many, many enthusiastic students and postdocs to share each square meter with. Again, probably the norm in most academic labs. In Australia we were spoiled: we had multiple rows of dedicated bench space for phage production, new state of the art purification equipment no one else was using. For phage purification, for example, I didn’t have to contend with much shared equipment, which meant when I left a machine, I knew no one would use it between my runs. I didn’t have to implement as many checks for cleanliness (ie. is there phage circulating in the AKTA from the last person’s run? Have to check before every run, which means a titre + extra waiting day before any new run). I also had a validated, calibrated endotoxin reader available that I could use, so I didn’t have to validate biochemical assays myself to be able to trust an endotoxin reading. Anyway to sum it up, having a bunch of empty space and pristine new machines no one else needs to use, plus a free human therapeutic quality control lab with professionals who maintain the equipment, is not really a reality in most academic labs…

Quality control personnel/facilities: As mentioned above, in Australia we had walking-distance access to a CAR-T therapy center to do our quality testing (and they were fine with us bringing phages in there!). We had a hospital micro lab down the hall that was happy to do our sterility testing. And we had a bioinformatician and sequencing team dedicated to helping us set up a QC pipeline for patient-ready phages. I’m sure these types of resources do exist here at Stanford (I’m new, so of course it takes time to identify and vet these resources). But so far I haven’t been able to convince the CAR-T centers here use their equipment for our phages, and haven’t yet attempted to see if the hospital lab will do our sterility testing. The students in this lab who did bioinformatics graduated just as I arrived, and my beginner Geneious & coding skills are hardly sufficient to make much progress fast. Sequencing facilities of course exist here, and we have a lab upstairs willing to do it for us, but turnaround time has been challenging, and setting up a pipeline for this that you know works for your purposes (e.g. how do we know we have all the reads we need to make a call about sample purity?) without dedicated bioinformatics support has been hard.

So this year for me has been a lesson in unseen advantages. Things about your current setup that you don’t think about, but are part of the reason for your success. When you have to fit in ‘set up a phage therapy pipeline’ between writing, mentoring students, and other research projects and meetings (no doubt the case for 99.9% of academic labs), phage therapy for patients becomes an uphill, almost unfathomable battle. It becomes hard to imagine how to do it safely. It makes me feel sheepish that I ever went around saying ‘an academic lab can do this’. We weren’t really a normal academic lab over at Westmead. It makes me even more grateful for getting to be a part of that over there, and makes me realize there’s a lot more unseen components to scope out and name before I can deliver a ‘phage therapy center kit’ on a platter and expect a normal academic lab to be able to run with it. I’m so grateful that this current experience let me learn that!

Of course, I’ve only been in this lab for 7 months. There’s a ton more I could still do to get a phage production system working. I certainly have the freedom to do it; a supportive PI and labmates. Stanford has a ton of resources, no doubt; I just probably need more time to figure out what and where they are, more time to meet the people who can help me surmount some of these hurdles, etc. And just more time to pause and articulate the needs and bottlenecks (this post is helping me finally do this!) so I can unlock what I need to solve them.

All in all, I have still made strides; I’ve started getting a phage biobank set up, establishing a sequencing pipeline, getting our phages sequenced, implementing a database of our phage stocks and batches, a batch labeling system, logbooks for our repetitive experimental work like titres and propagation runs. I’ve learned to use our AKTA Purifier and OH-chromatography (which works great!), figured out what it takes to clean it (and prove it’s clean) so it can be reused for new phages each time, and I’ve evaluated three endotoxin testing methods (to save you all the time, the best is what we were using in Australia; every phage lab should just buy an Endosafe® nexgen-PTS™ endotoxin reader and be done with it). Lastly, I’ve gotten to work with some truly amazing students here; it has been absolutely mind boggling to see how self-motivated, fearless and fast-learning Stanford students are! (They write and submit their own research papers and genbank sequences without help, learn how to use new equipment on weekends, and constantly make me question if we ARE going too slow with our progress…)

Overall, the experience of trying to do the same thing in two consecutive settings, each with their different strengths and limitations, has given me a ton of new perspective I didn’t have before. I know this will serve me going forward as I try to support others getting phage therapy off the ground in their institutions. This year I’ve felt at times like I’ve been spending 100% of my effort to get 10% of the way toward something I already achieved once, and this has made me question if I’ve been going about this the right way. Maybe instead of trying to copy what works in one setting, we should look at what we each do really well, and have unique access to, and build on that. Stanford has such cutting edge research going on, so much proximity to AI and machine learning experts, access to research funding, and access to incredible students. What could we do here, that we couldn’t do elsewhere? This is where my head is at lately. To be continued on where that leads!

Lessons in grant-writing

One goal I’m proud to have accomplished this year was to learn about grant writing — I got a TON of hands on experience with several grants, and feel like I can understand the anatomy of how they work in this country (at least for the NIH). I worked on 3 center-level (ie. ginormous multi-PI, multi-project, sometimes 600-page) NIH grants and helped respond to reviewer comments/resubmit a handful of R01 grants, which was enlightening too. I learned there’s a whole suite of rules about what makes a good one — more on these learnings in a future post! (And places like Stanford actually teach their students and staff courses on how to read between the lines and structure everything exactly right — not to mention giving labs full-time grant writing staff to help them and take care of the admin aspects… talk about unseen advantages!). Very grateful for the mentorship I’ve received from Paul on the grantwriting process, and for how much I was able to get exposed to in a short time. I also got to feel the pain of not getting two of them (so far). Ouch. It’s not fun, but it’s interesting to see the comments and the feedback you get — it’s way more detailed than I would have thought. I could see how you could build on this and how it would become fun…

Lessons in review writing

I am also proud of a review I submitted on phage therapy this year (it just got accepted today, yay!). My first ‘last-author’ paper, which feels pretty neat. Shoutout to the huge team of students and postdocs who each contributed parts of this (Saumel Perez, Grace Cullen, Zhiwei Li, Kevin Chen, Tejas Dharmaraj, and Tony Chang), to the collaborators who helped shape it (Gina Suh, Rob Lavigne, Sabrina Green, Jean-Paul Pirnay), to Paul Bollyky for instigating it, providing mentorship on how to manage the whole thing, and keeping us all motivated, and to Kevin Kim for driving it forward and doing the vast bulk of the work. And to Arya Khosravi for the support behind the scenes; offering help and edits and ideas.

Bollyky lab labmates! Some of the best people I have met. This was Halloween when we dressed up as minions… I still have not seen Minions or adjacent movies BUT turns out that wasn’t a requirement.

Lessons in podcasting

Lastly, I am SUPER proud that I answered my own nagging curiosity and got a podcast off the ground this year: Podovirus. I was both curious about the medium (would I like it as much as I like listening to podcasts? So far yes) and about the content (WHY don’t we have phage therapy yet? What is the hold up? This is driving the conversations and choice of guests so far). We’ve just released the 5th episode, with Sandro Sulakvelidze of Intralytix, and have 3 more that are recorded and mostly edited; almost ready to go. This has been in collaboration with Joe Campbell, who retired from NIAID this year, and he’s been fantastic to work with! It’s been a ton of work but I’ve loved exploring this new medium. More to come in the new year!

That’s all from me for now! I hope everyone has a wonderful holiday season, and I hope you each get and take a real break. I will be in Calgary with Jan and my family, relishing the fact that they are 2.5 hours away instead of 16, which was the real reason we moved our lives back across the pond after all.

~ Jessica

Jan’s year in reflection

So while Jess have been getting back into lab and grant work, what have I been up to? Well, I was going to upgrade Phage Directory but instead I got distracted. Now to be fair, this is basically every year, that’s why Phage Directory’s code is kind of in shambles, and lots of things are broken, and I make the promise that next year is the year I fix and rewrite all of Phage Directory.

My main distraction this year was everything AI. We even launched a new effort separate from Phage Directory to explore these ideas further, at labspace.ai, where we build experimental data and interface tools for data management, bioinformatics, and automatic Phage Directory and Capsid & Tail.

labspace.ai is a collection of experimental data management and research tools. Take a look — they’re all free to use!

Up until this year, a lot of stuff in C&T was manual — from cropping and resizing cover images to copying and pasting text into Mailchimp. This year, we built a lot of data flows to automate that for us. As you probably noticed, we also built a lot of tools that helps us tweet, summarize, and help us understand articles with science communication. Behind the scenes of Phage Directory we’ve changed almost everything; swapped out the API, upgraded the database, etc. which allows us to update the way we’re handling profiles, integrate next year’s Evergreen directly into Phage Directory, and a whole lot more. On the side of doing phage data, we’ve taken what we’ve learned from Phage Australia and built a bunch of tools for biobank data collection, which we haven’t rolled out yet, but includes neat features like a phage/host range heat map visualizer and Circos-like chord diagram system, to draw phage-host relationships (or co-authorship/collaboration relationships!). I’m even excited to add a better search mechanism, add better filters on Capsid, group special issues like Phage Picks, and be able to just read articles without the What’s New parts. Oh, and I’d absolutely love to build tools no one’s asked for, like a way to chat with Capsid & Tail 😉*.*

Since a lot of the upgrades have been on the side of code, it’s been really hard to show what’s going on — and if I wrote a post about it without the tools being fully available, they’re just kind of fruitless demos. Most of the systems are now in-place though, so next year we’ll definitely be making some sweeping changes around Capsid & Tail and Phage Directory.

Instead of announcing and promising a bunch of things next year, I’ll just be building those out over the next month or so, and hopefully you’ll just start noticing those changes throughout Phage Directory!

One thing we’re looking forward to next year is to just simplify-simplify-simplify and go back to the roots of Phage Directory — back to being a cozy close-knit community of phage researchers. And Evergreen. We’re definitely doing Evergreen again next year!

~ Jan & Jessica