OPIG Retreat 2023 | Oxford Protein Informatics Group

With the new academic year approaching, OPIG flew off to the rural paradise of Wilderhope Manor in sunny Shropshire for their annual group getaway. The goal of this retreat was assumed to be a mixture of team building, ‘conference-esque’ academic immersion, a reconnection with nature in the British countryside, and of course, a bit of fun. It is fair to say OPIG Retreat ‘23 delivered on all accounts, leaving the OPIGlets refreshed and ready for what the next year may bring.

Our journey began with an uneventful coach trip, with respect to the return journey (more on that later), up the motorway and round Birmingham on Tuesday afternoon. All parties arrived and those of us new to the experience soon understood the ‘retreat’ or reconnection with nature side of things as soon as we stepped off the bus. Trees, fields, cattle and farmyard manure surrounded us. The views from our new home were spectacular, as to be expected in Shropshire, and phone signal, as to be expected in Shropshire…was sparse.

In the military-like efficiency OPIG is famous for, we staked out our quarters: Fen’s nest, Hopes Dale, Wenlock Edge and Pudding Bowl being but a few of these names, shot-gunned our respective top bunks, deposited belongings and assembled in the cellar turned auditorium at 1740 hours for the start of the talks.

Guy got us started, guiding us gallantly through some of the more nefarious uses of artificial intelligence tools by those who seek to do harm or just want to watch the world burn. This was followed by the first of many excellent post talk discussions, with Garrett reflecting on the use, and misuse of fire, and the many parallels there. Steph had the second stint. The topic was ethnic diversity in biomedical research, from the academic community conducting, to the cohorts involved in clinical trials, the impacts and then the future of regulation. Intense discussion was followed up by takeaway Thai dinner in the great hall.

The second session of the evening was science heavy. Bora discussed a recent study on antibody humanness and stability (https://doi.org/10.1038/s41551-023-01079-1) that was of interest to many of us in the audience. Successively, Garrett kickstarted some discussion around small molecules by discussing how can molecules be better represented which translates into better models. We had a short break, then Lewis, a cup of coffee in hand, serious look on face, talked pollen. Tears, ‘bad’ antibodies, and parasitic worms all featured in a journey of one man’s silent struggle with hay fever and antihistamines. You had to be there. The evening was closed by Olly the elder – fellowship applications were on the agenda, and we were thoroughly treated to some excellent nuggets of advice, anecdotes, and more post-talk discussion.

Day one, nearly done. Bed was forbidden, as twenty past the hour of twenty-one ‘twas the time for enforced fun! An array of board games and then finally to bed. Now, this blog post has only three authors, so we cannot speak for the nighttime experiences of all OPIGlets. However, that first night led some of us to question whether the beautiful Manor of Wilderhope was indeed haunted. Howls of maniacal laughter drifted from the spiral staircase to our beds. Random nonsensical German and Portuguese words echoed in the darkness of some dorms and there was many a rushing of water and turning of taps acutely heard by those of us sleeping near the toilets. However, the pleasant evening temperatures and fresh country air were enjoyed by all, and we rose for day two ready for battle.

Breakfast. Then straight into the talks. Alissa presented on her recent blog post detailing how we can overcome limited data in ML approaches (https://www.blopig.com/blog/2023/06/machine-learning-strategies-to-overcome-limited-data-availability/ ), followed by Arun who talked about considering expressiveness of graph neural networks used in chemistry and molecular tasks. Nele delved deep into a diffusion-based method for all-atom antibody generation (https://arxiv.org/abs/2308.05027). In addition to going through the paper, this included some brief and useful notes on diffusion models, much appreciated by those of us still holding on to notions of semi-permeable membranes and glucose gradients.

After the break, Dan spoke about his recent work on TCR specificity and clustering models (https://www.biorxiv.org/content/10.1101/2023.08.04.551940v1), and then Carlos took the stage. Picture the following: Cows. A popular fast-food outlet. Block based flow diagrams and GPU optimisation. Many in the audience were stunned. Others, given the proximity to lunch and the delicious visuals, were just plain hungry.

Fabian followed, and it was colourful. At this point we should alert readers that it was not just the opinion of fellow OPIGlets and kudos up for grab in these talks. We had been informed by the organisers that there were several prizes on offer, voted for by the audience, and each related to a theme such as best storyteller, most entertaining, best haircut, and most colourful slides. This talk had been designed solely for the purpose of winning the latter, and the stall was set out from the start. The topic of protein flexibility was lost amidst a sea of colour that ensured when the dust settled, our man had his prize.

Quick break, then Martin talked about his recently released tool, ‘PoseBusters’ (https://arxiv.org/abs/2308.05777) and Leo gave conference feedback from his recent trip to ICML in Hawaii. Jay finished the morning talks by speaking about the history of sequence search and comparison. This talk had everything. We had colours, sequences, 1980’s GUI interfaces, celebrity photos, and BLAST. Unsurprisingly Jay took home the narrator prize.

Time for lunch, after which we all had the afternoon off. Some of us slept, others worked on our talks, while others went for a walk and soaked up some more nature. A precious few OPIGlets waited their turn to make dinner.

Cooking for over 30 hungry PhD and Post Docs, each with their own dietary requirements is no easy task, particularly in a small kitchen under the guidance of Olly and Teo, whilst following the printed-out, intricate, vegan-friendly recipes of our Scottish head chef who was no longer able to make the trip. However, both nights went rather well, and the food was delicious.

The evening talks on day 2 were of a similarly high standard. Henriette and Anna joined forces to tackle all things coeliac. Food, history, shopping, medicine, and immunology were all covered in a fantastic overview that prompted engaged discussion and many further questions. Matt finished off the evening discussing the power, potential, career benefits and daily difficulties of building and maintaining publicly available databases. Some of which you may be familiar with:

The day finished with a quiz led by occupants of The Cube (office 2.19), Lucy and Gemma. Questions started innocently enough with unsuspecting OPIGlets tested on their general knowledge, OPIG trivia and ability to identify famous cartoon pigs. As things started to get competitive, the quiz masters were met with mild heckling, requests for half points and disputes over answers. The following rounds gave way to more unconventional takes on the standard pub quiz with ‘Acronym or Wackronym’ (is this a real or fake acronym for a bioinformatics tool) and ‘Who Knows Whose Nose?’ (guess the OPIGlet member from an image of their nose – this one was surprisingly tricky!). Finally, to much amusement (to say the least) OPIGlets were tasked with guessing the correct hashtag to match an eclectic selection of Broncio’s tweets. Ultimately, team ‘Guy Nose Best’ emerged victorious.

Day 3 began with another joint talk on transformers in the field of proteins and antibodies with Ollie and Isaac. This included another excellent overview for those of us more familiar with the cars that become robots, and contained many useful links and papers that were furiously noted down by the audience. Benjie then talked about the oddities of PDB files and the many hidden gems to be found when searching through all things structural. Lucy took things up to eleven with analogue computing, discussing recent retro-technological advances that may change the way we build neural networks. Kate then told us about open-source pharma. This was an interesting concept in drug discovery that doesn’t get enough attention and challenges the traditional approach to making medicines. We had another lively discussion before taking a quick break.

Broncio restarted the final pre-lunch session. The topic again was open source, but now software (https://www.openbiosim.org/). The live demos went as smoothly as live coding demos can possibly go, and he finished within his allotted time. Gemma then talked sharks, specifically their strange immune systems and why this warranted OPIG investing in a shark tank. Colours were again widely used, and the visuals gave Fabian a run for his money. Alex closed the session talking about immunotherapy. Such was his enthusiasm for the subject that he ran over time, and everybody was late for lunch.

Walks, some indoor and outdoor games, and another successful dinner passed before the OPIGlets found themselves in the cellar for the final session. And what a session it was…

The penultimate two talks could in many ways be described as ‘postmodern.’ Lots of loosely linked ideas and themes, political messages, and bewildered onlookers. Alexi began by discussing a recent article on the R&D inefficiencies of big pharma. His excellent delivery veered into unknown areas of dark microbes, government research strategy and OPIG project design, with lots of audience participation. Teo then took the reins, and then took us further down the evening’s rabbit hole. Light, hue, colour scales, the beautiful visuals of ggplot and then a raucous finale which involved a debate on the true shade of Shrek which left the audience strewn across the complete colour spectrum of emotions.

The grand finale was Charlotte. Nobel prizes, Jerry Maguire, bicycles, and cryptic crossword puzzles were united by a common theme. Rather than give the game away perhaps we can leave readers with a clue:

Juggle in existence, Sir Van I came as rusty dude, producing progress (7, 6, 6)

The evening finished with prizes and giving of trophies. Many rounds of applause, then some more board games and deep meaningful conversation. A very relaxing end to a wonderful retreat. The next morning, we woke up, tidied up and packed up and all were onto the bus for 10am sharp. The less said about the 5-hour journey home the better…

Written by Alex, Gemma & Arun

Authors