Nina’s sister didn’t make it. A year old, she got sick on the train and never got better, dying in a dimly lit freight car. Babushka Nina still trembled thinking about her. Nina’s mom was Rh-negative, and her father was Rh-positive, so while the first pregnancy (Nina) went fine, during birth, the Rh-negative mom is exposed to the Rh-positive antigens from the first baby and develops antibodies against Rh-positive, as if it’s a virus or bacteria. For the second pregnancy, the mom’s anti-Rh-positive antibodies pass the placental barrier and attack the fetus’ red blood cells. While the second baby can be born, it dies due to Rhesus disease. Today, this is completely preventable with RhIG shot, costing about $370 on average in the US (if you’re lucky, could be $0 after insurance coverage), a small price to pay for a life.

After several weeks on the train, they traveled 2,012 miles (3,238 kilometers) to Bukhara, Uzbekistan. Her grandparents worked in textile factories to make ends meet. Nina remembers the delicious apricots. Her mother continued to work in the textile factory, while her father became a military driver and was concussed after a bad bombing. He lost his papers and fled to Kamchatka to avoid getting arrested.

After the war, the family (minus Nina’s father) returned to the Moscow region. There, Nina’s mother, Anastasia, faced an impossible choice that cost her life. Born in 1917, she was the same age as the Soviet Union itself. But while the Motherland grew, it created policies that killed mothers: outlawing abortion in 1936, after it was legal from 1920-1936. In her third pregnancy, knowing her child wouldn’t survive due to the Rh factor, she could either carry to term only to watch another child die or risk an illegal abortion and face death herself. Access to birth control would have saved her, but hormonal contraception wasn’t invented until the 1960s, 15 years too late. She got an infection and developed sepsis, normally treatable with penicillin, freely available in other countries. But in the Soviet Union, where corruption contaminated everyday life, the doctor demanded a 300 Ruble bribe by tomorrow to treat her.

Her grandparents scrambled to save their daughter. Her grandfather, a skilled repairman, took the express train to a neighboring district to work day and night at another factory. He earned the 300 Rubles, but came back a day too late. His daughter had already died. Angered beyond recognition, his toughened hands slammed the banknotes onto the floor where they lay worthless. A year of rent in Lenin’s useless faces littered the floor, staring back at eight-year-old Nina sitting in the corner, searing an image in her memory for the next eight decades. The image of money that could have saved her mother yesterday, but couldn’t bring her back today.

Nina worked as a radio engineer on satellites to send rockets into space in the secret town of Zheleznogorsk, Siberia for her adult life. In the 2010s, she took a computer class to learn to use a laptop and Skype to call her relatives in the US. At the beginning of her computer classes, she’d say to the computer mouse, “You and I are going to be friends today.” Thanks to her efforts, we talked regularly on Skype since 2012, until her passing in 2024.

Her stories live in my bones and flow in my veins to you, a reminder that the challenges of starting Seanome pale in comparison to the hardships she endured in her first decade of life. Watching her infant sister die on a refugee train, losing her mother to the failed Soviet system, and seeing her grandfather’s anguish over meaningless money, these moments shaped her life and echo through mine. I’m grateful to learn from Babushka Nina’s strength, so I can appreciate the privilege of building Seanome today.

Check out our longevity-db HuggingFace organization for all the uploaded datasets from the Hackathon. Amazingly, some of the datasets already have ~100 downloads after being online for only few weeks! How cool! It feels great that we’re already making an impact on the field with this work.

1st Place: Longevity-genie ($4k)

This team went for quality over quantity, and the judges noticed. We loved their clean, well documented repositories (both code and data!), and importantly, these datasets also loaded with the HuggingFace datasets API! Amazingly, this team also won another hackathon in Berlin the same weekend! I’m so impressed!

Team members

• Newton Winter (in the UK!)

• Brandon van Houten

• Anton Kulaga (in the UK!)

skeletal-muscle-atlas

As muscle is a clear biomarker of aging, with grip strength and ease of building muscle declining with age, we need to understand what’s happening at the molecular level so we can design effective interventions to prevent muscle loss.

aging-fly-cell-atlas

Aging of a whole organism across life is a HUGELY valuable dataset! However, obtaining such a valuable dataset from humans is extremely labor-intensive and for now, we’ll settle for using the fruit fly (Drosophila) as a model. While we’re not flies, we share many similar basic mechanisms like cell division, which tends to break down with age. By making this dataset accessible to the broader AI/ML community, we can learn how aging is a systemic process that affects whole body systems, which systems are most affected, and where can we best intervene.

2nd Place: CellVPA ($2k)

CellVPA did a great job of preparing the data, having beautiful READMEs, and showing their work with their processing scripts. Thank you, CellVPA!

Team Members

• Pooja Dolas

• Venkatachalam Subramanian Periya Subbu

human-optic-nerve-fibroblasts-snRNAseq

The eye is extremely susceptible to age-related degeneration, and its senescence can lead to glaucoma which is the leading cause of irreversible blindness. A single-cell RNA-seq dataset describing the intricate cell types of the eye is crucial to prevent eye aging.

human-muscle-aging-atlas-snRNAseq

Aging muscle is a popular dataset! This team also used the same Human Skeletal Muscle cell atlas as longevity-genie, and I think it’s great to have the dataset uploaded multiple times. This team did a nice job of giving instructions to ML engineers for how to use the data.

mouse-muscle-aging-atlas-snRNAseq

More aging muscle! This time in mouse, which means that the data is much more controlled, and researchers can really dig into the aging signal specifically, and not have as many confounding variables as in human like lifestyle choices.

3rd Place: Florian De Rop ($1k)

3rd place went to a team of one! Florian had a major push with great data processing scripts, and data that loaded on the HuggingFace API. Way to go!

terekhova2023

This dataset from blood of individuals aged 25-85 will have the easiest application towards diagnostics and biomarkers for aging. As peripheral blood mononuclear cells (PBMCs) are the easiest tissue to biopsy, this dataset will be immensely useful for researchers who want to find genes that change across age, and quantitatively measuring their difference after a longevity intervention.

riken2018

Also from blood, this dataset features samples from supercentarians (ages 110+!) and younger controls. What an amazing opportunity to learn from what are some positive biomarkers of aging for the healthy elderly! I’m really excited for what people do with this dataset.

All high-quality datasets

Here are all the HuggingFace datasets that the judges rated favorably, regardless of team. Enjoy training models on them!

• https://huggingface.co/datasets/longevity-db/GSE120180

• https://huggingface.co/datasets/longevity-db/aging-fly-cell-atlas

• https://huggingface.co/datasets/longevity-db/human-muscle-aging-atlas-snRNAseq

• https://huggingface.co/datasets/longevity-db/human-optic-nerve-fibroblasts-snRNAseq

• https://huggingface.co/datasets/longevity-db/mouse-muscle-aging-atlas-snRNAseq

• https://huggingface.co/datasets/longevity-db/pan-cancer-nuclei-seg

• https://huggingface.co/datasets/longevity-db/riken2018

• https://huggingface.co/datasets/longevity-db/skeletal-muscle-atlas

• https://huggingface.co/datasets/longevity-db/terekhova2023

At Seanome, we’re building open-source tools that figure out the “jobs” of proteins in the cell, unlocking the forgotten 99% of life on Earth. The vast majority (far greater than 99%) of proteins have unknown functions, and we’re changing that by building Kmerseek to identify more novel protein functions, faster.

Being part of the nf-core community is key to Seanome’s success because we want to engage with the global community of tool and pipeline builders. We are developing our own tool for annotating unknown proteins (Kmerseek) but to do that, we need to (1) benchmark it against the competition to demonstrate Kmerseek can find things other tools can’t and (2) empower OTHERS to use it as easily as possible! Hence, we are building nf-core/proteinannotator, a unified Nextflow pipeline to characterize “mystery” proteins, aka proteins of unknown function from their sequence data.

To engage with the global community of pipeline builders, I attended the 2025 Boston Nextflow Summit, a 4-day event from May 13-16. It was divided into 2 days of training or hackathon, and 2 days of Summit talks.

It was fun reconnecting with people I hadn’t seen since the last Nextflow event I attended — back in 2019 when it was called “Nextflow Camp!”

For the first two days, I led a hackathon group for nf-core/proteinannotator. The goal is for someone who sequenced a cool new creature to use our pipeline to figure out what their proteins do!

We made a bunch of progress: hackathon attendees added 4 tools to the pipeline! Key ones were mTM-align, MMseqs2, HMMer, and of course, BLASTp. Thanks to Chase, Addy, and Jessica for helping! Integration of the tools into the pipeline is still in progress and help is welcome!

Seqera was kind enough to give me a speaking slot where I talked about Seanome and how we’re building the best protein annotator in the world! It was fun to do this back-to-back with the hackathon and update the community on our progress. Check out the recording of the talk on YouTube.

I’m excited for more progress on nf-core/proteinannotator! This week I merged a long-standing pull request to add InterProScan to the pipeline, which I started at the March 2025 nf-core Hackathon.

If you want to help out with Seanome, the best place to contribute by picking an issue on the nf-core/proteinannotator GitHub repo and get coding!

Last weekend, Seanome hosted a Longevity x AI Hackathon with Frontier Tower (Thanks Elliot Roth!). I was part of a team (along with JJ Hales and Karl Pfleger) that won $10k from Foresight Institute’s Vision Weekend in December 2024. We were the Longevity/Biotech team competing against Neurotech, Nanotech, AI Safety, Space, Existential Hope for $10k to accelerate our field. Our team won with the premise of breaking down barriers for AI/ML in aging: getting molecular profiling datasets of biological aging (especially those curated by Karl and Francesco Neri in the Aging Biotech database!) into the hands of AI/ML practitioners. Originally, we wanted a Kaggle Competition, but now, to run a competition, you need $100k (!) and thus we were priced out with our measly $10k. We brainstormed and came up with a hackathon!

For this hackathon, our goal was to upload as many aging datasets as possible to HuggingFace, a popular AI/ML platform. Of course, someone asked …

“Uploading datasets? Isn’t that trivial?”

I responded, “… I think you underestimate the complexity of biological datasets.”

Currently, working with a biological dataset requires know-how for (1) locating the data, (2) understanding its meaning (e.g. row and column definitions), and (3) its potential uses (e.g. does it indicate gene activation? Can it identify pathways?). With Karl’s curation, (1) is somewhat solved, but 2-3 are still challenging. You need to read the paper and dig into supplementary files to know what is going on. Datasets on cellxgene (”cell-by-gene”) are MUCH easier to use as they are formatted and conform to a schema, but still take know-how (I’ve heard it can take engineering teams ~months to ingest them!). But most biological data is NOT on cellxgene, but on National Center for Biotechnology Information’s (NCBI) Gene Expression Omnibus (GEO). If not, maybe the author made the data available as a supplementary table, if you’re lucky! If you’re unlucky, you need to email the author to get your hands on the data :(

Datasets on cellxgene still need formatting to get them HuggingFace ready. See seanome/2025-longevity-x-ai-hackathon for example data cleaning notebooks.

I had low expectations: more attendees than me and the organizers, and at least 1 dataset uploaded. We crushed them! Over two days, we had:

• 16 participants! (The winning group was split across two continents: SF to London, wow!)

• 36 longevity molecular profiling datasets uploaded to HuggingFace!

• 166.6 GB uploaded to HuggingFace!

• $7k awarded to 1st ($4k), 2nd ($2k), and 3rd ($1k) place teams!

Thank you to our judges Zoe Isabel Senón, JJ Hales, William Poole, and Aabha Morey for lending their expertise and helping select our winners!

Things I’d do differently for the next hackathon

This is my first time running a cash prize hackathon, and it’s different from the nf-core hackathons I’ve run before.

Open by default

Our hackathon goal was to contribute longevity datasets to HuggingFace. However, to advance the field, we also need to share the code that created the uploaded data. It’s not enough to share the data; we also want to share the code, as users will want to see how it was prepared, and it will help others in the future as examples of cleaning up the data! Later, we added the requirement to share the processing code, which wasn’t compatible with all participants, and we can be clearer about this in the future.

I’d be clearer about another requirement: for the HuggingFace dataset repository to link to the GitHub repository and vice versa, to help future users understand the data processing and for coders to see its HuggingFace format.

Invite more people!

We budgeted food for ~40 people and in my naiveté, I thought inviting 40 meant 40 would come! But no…. San Francisco has a LOT going on, and I was told to expect 50% (!) to show up. I ignored them and kept the total count at 40, asking people to decline if they couldn’t make it anymore. We had 14/40 (35%) in-person participants, which is on par for the 30-60% I’ve heard for events in SF. If I ran this again, I’d have a 100-person guest list and advertise more widely and earlier.

Teams

I don’t mind if you code in a room by yourself outside of a hackathon. HOWEVER, the point of a hackathon is to meet other people! We had a few going solo, and one big (5-person) team, and 2-3 in a team is a nice sweet spot, because you’re still collaborating but not too many that the coordination cost gets too high. Next time, I would state directly in the event text that you’re required to work in a team of 2-3, and we won’t judge submissions from a single individual.

Overall, I’m grateful to the opportunity from Foresight Institute to meet the lovely hackathon attendees and people of the Frontier Tower! Looking forward to more events there!

I talked about my 15+ year journey from a double-major in Biological Engineering and Mathematics at MIT (since they didn’t have a computational biology major yet! I had to invent my own), being lucky enough to work in Sean Eddy’s lab as a summer intern at Janelia Farm Research Campus, working in Jill Mesirov’s lab at the Broad Institute, doing a Bioinformatics Master’s degree at UCSC, then a Bioinformatics PhD at UCSD, getting to be part of the early data science team at Chan Zuckerberg Biohub, a stint consulting for Arcadia Science, then learning about drug development at BridgeBio, and finally starting Seanome last year.

A big reason WHY I started Seanome is Botryllus schlosseri, the coolest organism you’ve never heard of. My collaborator, Prof. Ayelet Voskoboynik (also on Seanome’s scientific advisory board!) introduced me to Botryllus Histocompatibility Factor (BHF), a gene that is very similar to our acute organ transplant rejection genes aka histocompatibility genes. AND YET, this gene has no sequence similarity to any human genes! I became obsessed with this and wanted to figure out a flexible way to find similarities between really different genes, and lead to writing the prototype (see an example run here!)

I’m also really excited to reconnect with the nf-core open-source community, which is a lovely group of people helping each other out by making best-practices pipelines in genomics and beyond. I’ve contributed to nf-core/rnaseq, nf-core/scrnaseq, wrote my own nf-core/kmermaid, and am now leading development for nf-core/proteinannotator. We made a bunch of progress on proteinannotator at the May 2025 nf-core Hackathon in Boston at the Nextflow Summit — stay tuned for progress!

🎧 Listen here: Episode 49 — Olga Botvinnik on marine genomics, open science & Seanome

Last week, I gave the weekly talk at Stanford’s Evolgenome Seminar Series, highlighting cutting-edge work in evolutionary genomics. I was nervous, as the series featured heroes of mine like Richard Durbin (whose textbook I studied), Katie Pollard (whose human accelerated regions I used in grad school), and Alex Pollen (whose brain single-cell RNA-seq paper and data I used in grad school). It felt like I had achieved a significant milestone with this talk.

My presentation, “Unlocking Hidden Protein Functions with a Biochemically Informed Annotation Strategy,” showcased how we’re using hydrophobic-polar k-mers and laptop-scale computation to illuminate the uncharacterized protein universe. I explained how our approach identifies potential functions in protein regions that resist analysis by conventional tools like BLAST and HMMer, and sophisticated AI approaches like FoldSeek/AlphaFold. I shared preliminary results annotating a transplant rejection immune gene in our closest non-vertebrate relative. As part of our mission to decode the “mystery genes” of marine organisms, I briefly touched on our work with Arctic clams and their neurotoxin tolerance mechanisms.

I used Slido to engage the audience with interactive questions throughout the talk, which was fun! We collaborated on questions like "How many human proteins are of unknown function?," "How many GPU years were required to train AlphaFold3?," and my favorite, a word cloud of “What protein domains do you think are in a marine transplant rejection immune gene?” This made the talk feel more integrated with everyone. The Stanford researchers asked thoughtful questions about how our computational approaches could filter the vast protein sequences search space to manageable sets for deeper analysis.

Thanks to the invitation from Stanford’s Center for Computational, Evolutionary and Human Genomics. I’m excited for future collaborations to democratize marine genomics and unlock innovations in the forgotten 99% of ocean biodiversity!

Forgotten 99% of Life on Earth: The Quadrillion-Scale Challenge

AI in biology is overhyped because, unlike natural language, where every document has been observed, not every biological sequence has been seen, and thus is not in the training data.

Imagine trying to understand all human language by studying Russian math textbooks. That's what we're doing with protein language models today. The input data is biased towards one source: human. Even our best AI tools like AlphaFold, while a significant advancement, were trained on a tiny, biased fraction of nature's molecular library. Of the estimated 14 quadrillion proteins1 in nature2, we have three-dimensional structures for only ~67,0003 of them — and a big chunk are from one species, human!

For the remaining ~2.2 billion species on the planet, their proteins are written in a language that we barely understand, and we need tools to decode them. We need to look beyond what we've seen to build better translators. That's why Seanome is diving into the ocean: Earth's greatest laboratory of molecular innovation.

Why Ocean Animals are the Ideal Testing Ground

The ocean is home to animals with unique adaptations that can’t exist on land, and has given us one transformative technology — green fluorescent protein (GFP) from glowing jellyfish. Without GFP, we wouldn’t be able to develop COVID vaccines4 or create “brainbows” to study the connections of the brain’s 100 billion neurons5. A wealth of potential applications in human health and longevity swims in our seas. These include unfreezable fish whose antifreeze proteins could preserve organs indefinitely, immortal jellyfish whose reversal of aging could transform what we think of a “normal” lifespan, and Seanome’s pilot project on Arctic clams whose neurotoxin tolerance could cure neurodegenerative disease.

The extraordinary biodiversity of marine animals — representing 76% of animal life6 and 43 billion7 genes8 — created fast-evolving immune systems to battle 100 million viruses per teaspoon of seawater9. This rich ecosystem is nature’s most diverse testing ground and offers a potential gold mine of novel therapeutics (including antivirals, antibiotics, antifungals) and an ideal environment for building better protein analysis tools. If we can decode the ocean’s molecular mysteries, we can understand anything.

The Forgotten 99% Will Transform Our Future

Breakthroughs like Ozempic (derived from lizard venom) and anti-HIV and anti-cancer compounds from a single innovative organism, ancient sea sponges, give us a glimpse of what’s possible. These discoveries come from a handful of species representing a tiny fraction of Earth’s biodiversity. We’ve only begun to explore nature’s 4 billion years of experiments.

At Seanome, we believe nature has solved quadrillions of molecular problems, but we lack the keys to access the doors they are behind. We’re building those keys. By creating tools to better understand biology, we enable access to nature’s solutions to disease that were previously invisible.

By focusing on the overlooked 99% of animals, we’re empowering researchers to make discoveries that could transform human health, longevity, climate, agriculture, and more.

Countless discoveries await us, hidden in the genomes of the forgotten 99% of life on Earth. Seanome is committed to illuminating this path — one protein at a time.

At the Plant and Animal Genomes conference (see our recap), I had six days of note-taking to perfect my Notion templates. Let’s dive in!

Iterated on the best way to take notes for an external talk

I love taking detailed notes to pay attention and retain information from the talk. I’m a visual person, so having the slide pictures is immensely useful if I can get them. I’m the annoying one in front of you at the conference taking pictures of every slide.

For this conference, I set PAG32 as a project and added pages to the related “Meetings” database, using the “External Talk” template. See below for the full PAG32 project page and all relations.

Initially, the template was simple, but it became more powerful by Day 6. I refined on my note-taking style throughout the conference, taking dozens of notes to figure out the best method for me. This was a good reminder that nothing is “perfect” the first time, but taking action will help you understand what is or isn’t useful.

Initially, I just did notes and a picture

The simplest way to take notes was to use slide headers, add bullet points for the text, and a picture below. It was simple and it worked.

Switched to 2-column view

On Day 3, I realized the slide pictures took up a lot of vertical space. So, I started using the two-column arrangement in Notion for side-by-side annotations. To make it easy, I made a Notion Button (below).

The button was easy to use, and took up less vertical space, which was my goal! Here’s an example of slide note-taking:

This was nice, but because Notion is organized by “blocks” and the title, notes, and picture were in different “blocks,” it was difficult to reorder the slides if I mixed up the order while taking notes during speedy talks. In Notion, I could make a single block as a toggle or as a callout. Toggles hide extraneous information, but I wanted to see the photos and notes for each slide. Callouts keep the nested blocks together and add nice subtle visual box around the content. Here, I think callouts are more useful because the contents are visible in the whole page for easy scanning, while toggles would require clicking to open the notes for each slide.

Switched to callout + two-column view

By Day 5, I chose a callout to keep all photos for one slide in one place (for animations), save space, make each section movable, and let the slide title use the full page width. To get the two-column view in the callout, do some hacky Notion-ing: create a “new page” within that callout, make the two-column layout in this nested page, navigate back to the main page, and turn it into “text.” See this video for a demonstration.

Below is an example section of slide notes, including a couple of slide photos to capture the animations by the presenter.

Hope you enjoyed that deep dive into conference note-taking with Notion. Here’s the Notion template for the final version.

Here’s the gallery from one session! It turned out nicely.

Hello! I got back from the Plant and Animal Genomes 32, a week-long conference on — you guessed it — plant and animal genomes! The focus is on feeding the world through more nutritious crops and resource-efficient agriculture, but there’s more basic science as well.

I had a talk and poster (PDF links on Figshare) and got great feedback on our methods, potential users, and tools to benchmark against.

Talks I liked

There were many talks, but here are a few sessions I enjoyed:

• Day 1 (Friday):

  • ORG.one Genomes for Conservation of Endangered Species had interesting talks, including training undergraduates in biodiversity genomics, reintroducing the little blue macaw (like the “RIO” movie) into the wild, sequencing cute olinguitos in Ecuador, and conservation genome assembly in the Red Sea.

• Day 2 (Saturday):

  • Aquaculture discussed NCBI’s resources for annotating genomes with their new “datasets” tool and single-cell RNA-seq of catfish spleen (B cell development).

  • I appreciated the discussion of how Gregor Mendel’s pea plant experiments were the best experiments ever done in the Comparative Genomics session by Shifeng Cheng (AGIS, CAAS).

• Day 3 (Sunday):

  • The Computational Genome Decoding was kind of spicy, with every speaker claiming their method was better than the previous one.

  • I love learning about the protein sequences and annotations in the UniProt Database!

• Day 4 (Monday)

  • PacBio’s HiFi session had great talks about the additional information gained from high-quality genomes for fish and the ambitious Earth Biogenome Project to sequence all planet biodiversity.

  • PacBio had a great party that night!

• Day 5 (Tuesday)

  • By day 5 I was drained, but my talk was on day 5, so I still had to bring the energy!

  • The Plenary session about creating Vitamin D-rich tomatoes to supplement diets was cool! Apparently tomatoes already make a compound similar to Vitamin D, so the engineering wasn’t too bad.

  • The Bioinformatics session featured the legendary Sergey Koren presenting recent advances in full telomere-to-telomere human genomes and their application to other organisms. I appreciated the cautionary tale of sketching with only canonical DNA k-mers, leading to uncovered k-mer deserts in the genome.

  • The Proteomics session had talks about comparative eggshell development in chickens/ducks/pigeons/parrots and making less allergenic gluten in wheat through molecular dynamics protein structure simulations.

• Day 6 (Wednesday)

  • The plenary session on scalable plant genome editing discussed the GMO label and how manufacturers avoid engineered food because consumers want the “butterfly” non-GMO label. I asked why the industry doesn’t brand precision-engineered food as higher nutrition, no pesticides, and more flavor, (because that’s the reason I buy organic and local, if possible) and the speaker said consumers can’t tell the difference between an edit and a transgene, but I think they can. It’s fixing a typo vs inserting a whole paragraph. And we just had a worldwide education in mRNA vaccines! Maybe I’m naive, but I think it’s possible.

  • Wildlife Genomics had cool talks about the woolly mammoth genome, complex genomic structure in North African foxes (including the super-cute fennec fox), sequencing honeybees from Swedish museums to study species richness declines, genomics of sadly collapsing bird populations in Hawai’i, and the origins of sulfur-adapted fish in Mexico.

What I would have done differently

This was my first conference going completely solo, without fellow group members or coworkers, and had I known about it, I think I would have benefitted from PAG Mentorship Program. With over 3,000 attendees, it’s easy to get lost in the crowd at PAG! The mentorship program pairs first-timers at PAG with experienced attendees. They meet a few times on Zoom beforehand, have a breakfast at the meeting, and stay in touch throughout the event. I might have seen it when I was signing up, but arrogantly thought “oh I’ve been to conferences before.” I thought it was for people who hadn’t been to many conferences and I felt like I was fine. However, since I was going solo, I didn’t know many people, and I would have benefited from joining a group.

I learned about the program from the organizer of the PAG mentorship program, Mónica Muñoz Torres (LinkedIn) at lunch on the last day, who espoused its benefits! It made me think it would have been helpful to join a group like that.

Overall

It was a great conference, and I hope to be back at PAG33 next year with new results!

Next week, I’m speaking at the Plant and Animal Genomes (PAG) conference. I’m nervous because it’s my first conference talk as “Seanome Founder,” which feels like a big milestone!

I wanted to share how I prepare for talks. When I’m at my best, I fill out this Presentation Prep Notion Template for important talks. This template forces me to consider the audience, duration, number of slides, and outline. I wanted to share it, in case it helps anyone.

Let’s work through this template together in preparation for my talk!

Audience Questions

Below are the audience questions:

• Audience: ___

• Who are they?

• What do they know?

• What don’t they know?

• What do they care about?

• What don’t they care about?

Here’s how it looks in the Notion template:

”Who is the audience?” was simple — “Plant and Animal Genomes Conference Attendees.” That gave me a starting point to fill out the more in-depth questions. Once I thought through my answers, I asked Claude to fill out these questions using my Seanome Project in Claude with all my executive summaries, pitches, etc. Note: I wouldn’t have benefited from using Claude if I hadn’t made my brain work first!

Claude had interesting perspectives. I struggled to identify what my audience doesn’t care about, and it helped clarify what I can cut from the talk.

Here are my answers below:

Duration Questions

You’re usually given a specific duration for your talk, so this question is mostly out of your control. However, it’s helpful to consider how many slides you can present in the allotted time, as that will determine the pacing and per-slide density. Never have more slides than minutes. It’s not feasible for your audience to absorb information at that pace. Respect your audience!

• How much time? ___ minutes

  • Never have more slides than minutes of time

• How many slides? ___ slides

The number of slides depends on the type of presentation

• Lecture? → ~1 slide/minute

• Discussion? ~1 slide/2-5 minutes

  • [Note: this is a personal note from a previous presentation] e.g. recent Long Read Seq presentation: got through 15 slides in 60 minutes (including title and outline)

Here’s how it looks in the Notion template:

The pacing depends on your audience and venue. In academia, I was taught “1 minute = 1 slide” and “1 figure = 1 slide.” However, at Bridge Bio, slide decks were denser, with multiple figures (see Bridge Bio’s Investor Deck for November/December 2024). The slides drove discussion, rather than being a one-way transmission of information, and it could take up to 5 minutes per slide. I had to rethink my slide-making strategy at Bridge.

Back to the PAG talk prep. Here’s what my answers looked like:

Outline Questions

If you’re like me, you’re wondering, “but we haven’t made any slides yet!” It’s okay, we’ll get there.

This is where it matters. I struggle with writing outlines. I don’t know if it’s my ADHD, but making an outline feels unnecessary. Why can’t I just start making slides?! But as the note below says, I tend to jump into the “fun part” of the presentation, aka the data analysis, and don’t give enough context for my audience to understand WHY I’m presenting what I’m showing. They end up confused and frustrated, and I don’t get useful feedback on the work.

That’s why I created this template — to save myself from myself. I found somewhere that great presentations have a breakdown of 30-50% Background/Introduction, 40-60% “Main” Data/Methods/Results, and 10% Future Work and Conclusions. I forget where I got these numbers, so please comment if you know. Keeping these rough numbers in mind protects my audience from my data-heavy presentations. This forces me to pause and consider the audience and what information they need to understand my point.

This breakdown is helpful for me. It forces me to think about what I show if I can ONLY include 4 data/methods slides. I can’t dump a bunch of information on my audience and expect them to understand.

Honestly, I returned to my Seanome Project and had Claude draft an outline because it felt overwhelming to start. After some tweaking, here is my current outline.

15-slide outline

1. Title: Unlocking Hidden Protein Functions with a Biochemically Informed Annotation Strategy

2. Outline

3. Background: What is Seanome

4. Challenge: Scale of lack of annotated proteins

5. Technical Problem: Where current tools fail: Functional Homologs with low sequence similarity (BCL2/Ced9 example)

6. But what about AI?!??!

7. But what about AI?!??!: Part 1/3: AI is only as good as its training data

8. But what about AI?!??!: Part 2/3: Disordered protein regions aren’t represented in PDB

9. But what about AI?!??!: Part 3/3: AI is expensive and slow to integrate new data

10. Problem Introduction: Opportunity in Ocean Biodiversity

11. Methods: Biochemical data augmentation k-mers

12. Methods: Reduced amino acid alphabets

13. Case Study: Background on Botryllus schlosseri

14. Example/Case Study: Botryllus Histocompatibility Factor (BHF)

15. Conclusions

16. Future Work

I’ll turn those numbered lists into H3s (Command-Option-3 for the win!), and bullet point the sub-points. Here’s a preview of the first few slides. The title is the easiest one :)

I used Claude to flesh out the points because I tend to write a few keywords, and Claude will take that starting point to be more verbose, which is helpful. Most AIs are too wordy, which I find annoying, but here, it’s useful to be chatty.

It’s annoying because Notion headings don’t have automatic numbering, unlike LaTeX, so I need to manually change the heading numbers. I like the H3 because it makes each section feel distinct. Otherwise, it seems like a long list, making it hard to focus on one section.

I now start working on the slides! As I work, I notice logical gaps and extra sections to add. For example, I started with “But what about AI?!??!” as a single slide, but I kept jamming in too many figures onto one slide and realized it needed more.

Lately, I’ve been writing for an investor/philanthropic audience, which is very high-level, with almost no details. This presentation is a big change for me, since a scientific audience cares deeply about details and is skeptical of unsubstantiated statements!

I can try a few things to practice the talk. For past talks, I’ve recorded a single-person Zoom call and cringingly watched my presentation. I’ve heard of using Claude as a presentation coach by uploading an mp3 of my presentation and a pdf of the slides, so I’ll try that, too.

Now, back to preparing for this talk. Wish me luck!

Thanks to Claude and Lex for editing!

Note: The next post will be delayed, as PAG 32 (January 10th - 15th) overlaps with next week’s schedule of posting on January 14th. The next post will be an overview of PAG and will be posted on Thursday, January 16th.

To me, biology is a spiritual experience. I had a stark out-of-body experience in high-school biology when learning about ribosomes. I looked at my hand taking the notes about ribosomes, and realized that the ribosomes in my hand were translating mRNA to protein in this very moment. Like astronauts seeing the blue marble of Earth from space, I had an overview effect, a pseudospiritual experience realizing that biology is not only all around us, but within us all.

To communicate the depth of Seanome’s mission, here is our first company value:

1. Unity is Biodiversity

We are all connected through our molecules — the DNA, amino acids, and metabolites shared across all living things. While traditional science segments and separates, the ocean contains four billion years of evolutionary wisdom, and the origins of our deepest connections. From innovative body plans to ancient immune systems, by studying one organism, we illuminate the beauty within us all. We find strength in these connections, bringing together diverse perspectives to unlock nature's most profound secrets.

This value is 100% stolen from Millennium Dance Complex, aka “the dance studio where stars are born.” All major artists like Britney Spears, Usher, Justins Timberlake & Bieber, Pink, Katie Perry, and more, come to Millennium to train their bodies and minds to retain and perform choreography. Millennium’s iconic wall showcases their motto, “UNITY IS DIVERSITY.” They intentionally changed it from “UNITY IN DIVERSITY”, because the “IS” makes a stronger statement, that unity is defined by diverse parts making a greater whole.

Above is my favorite choreographer at Millennium, Galen Hooks and her phenomenal RIVER piece. If you haven’t seen it, go watch now, then watch the inside look, and then binge some more Galen Hooks choreography (recommend Bury a Friend, Baiana, and Love on the Brain) before coming back. I’ll wait.

Why can we get chills from watching dancers on a screen from a different time and space? We are affected because of our shared humanity, feelings, and spirit, which transcend time and space.

As a dancer, I believe we are all dancers: we all love to move our bodies to music. Rites of passage across world cultures such as weddings, coming-of-age, and funerals, wouldn’t include dance if it didn’t deeply affect us to our core. Our unity as humans is inherent — we wouldn’t love to dance if we didn’t find connection in it. And we wouldn’t be so affected by watching dancers if we couldn’t feel the same emotions they are conveying through movement.

Millennium posted on Instagram about “Unity in Diversity” (below), where they acknowledge we need all people to build the future we want to see. Unity is not just about tolerance of differences but seeing ourselves as connected to each other as people, as humanity. The more we accept our connections, the easier it is to love, to forgive, and see the good in one another.

https://www.instagram.com/mdcdance/p/BHkNamIAgBF/

* Unity in Diversity *

is a concept of "unity without uniformity and diversity without fragmentation" that shifts focus from unity based on a mere tolerance of physical, cultural, linguistic, social, religious, political, ideological and/or psychological differences towards a more complex unity based on an understanding that difference enriches human interactions.

#LOVE

Similarly, I believe we are all scientists, innately curious about the living world. We watch birds, grow plants, and immerse ourselves in nature to relax. Parents see nature’s wisdom unfold before their eyes, astonished by how a helpless infant grows into their own person with unique hopes and aspirations. These transformations remind us of our deep connection to nature.

When we accept that “Unity is Biodiversity,” we acknowledge that biodiversity unifies nature. “Unity in Biodiversity” suggests that we are unified despite our differences, while “Unity is Biodiversity” suggests we are unified because of our differences and the interactions they create. The plants, animals, and microbes interacting in the ocean are what make the ocean the ocean — otherwise it would be a big, lifeless basin of water. The beauty of biology is that interactions between species are only possible because of our molecular commonality.

We are connected to all living things through our molecular machines. Trees have proteins, fish have carbohydrates, and mushrooms have DNA. No matter how large or small a living thing is, we are all built from the same building blocks, and I think that is so profoundly beautiful.

Learning how a cell works in one organism makes us better understand the cells in ourselves. By studying how Arctic Clams tolerate neurotoxins, we hope to cure human neurodegenerative disease. The connectedness of life flows through our shared biology, bringing us closer to nature and each other.

At Seanome, we build open-source genomic tools to democratize discovery to anyone with an internet connection. By connecting researchers worldwide through their common love of biology, we enable scientists to reveal hidden connections across species — from Arctic clams to humans. Each discovery reveals another layer of molecular choreography, each species expressing its part in nature's ancient dance.

Thanks to Claude for editing.

Quick note — We will be taking time off for the holidays, and will be back on Jan 7th, 2025.

At Seanome, we believe evolution has already solved trillions of molecular problems. Sometimes, these ancient solutions surface from unexpected waters, such as a son’s quest to cure his father’s rare genetic disease leading him to the frigid depths of the Arctic Ocean.

Max Glanz witnessed nerve damage of unknown origin (idiopathic neuropathy) set his father's hands ablaze with relentless pain. A scrappy kid from Boston, Max couldn't sit idly by. Instead, he set out on an ambitious mission to cure his father. Along the way, he discovered an ancient healer dwelling in Arctic waters: the surf clam. His journey, captured in the documentary "My Last Nerve" (2024), reveals how nature's overlooked creatures, shaped by billions of years of evolution, have already mastered challenges that have stumped modern medicine.

At its core, neuropathy stems from a cellular betrayal — when a crucial enzyme meant to build protective compounds instead creates toxic ones that slowly destroy neurons, leading to loss of sensation and severe disability. While current treatments can slow this process¹, they cannot stop it. But here, in the intersection of human suffering and marine wisdom, lies a profound revelation: spisulosine, the very molecule that devastates humans, is naturally present in Arctic surf clams (Spisula polynyma)². Since the clams live with spisulosine, they must have a use for it and a way to degrade it. By developing open-source tools to mine their genome for this ancient adaptation, we aim to transform the clam’s evolutionary innovation into hope for treatment.

The connection between Arctic clams and neuropathy while looking for novel antitumoral compounds for cancer treatment. Spisulosine from Arctic clams upended current understanding of cell death by using a completely novel mechanism, and exhibited very high potency³. As spisulosine killed cells so effectively, it was immediately put to use for cancer treatment⁴⁵. Unfortunately, it didn’t pass Phase I clinical trials due to neurotoxicity leading to neuropathy⁶⁷⁸⁹ — the same symptoms as Max’s dad — hinting at the connection between clams and neuropathy. Not only does this compound affect cell death and neurons, but its broader molecular family (sphingolipids) is biomarkers of diseases ranging from diabetes to stroke¹¹.

This discovery reveals a profound truth: marine organisms, particularly long-lived species like clams¹², have mastered cellular existence in ways we are only beginning to understand. Their ability to precisely orchestrate cell death and renewal — a crucial aspect of longevity¹³ — demonstrates how discoveries from the ocean could uncover deep biological truths that span the entire landscape of life itself. Curing neurodegeneration, aging, diabetes, stroke are mere glimpses of the potential in biodiversity.

Max’s mission to cure his father drove him to the intersection of human health and ocean biodiversity. At Seanome, we believe this is just the beginning. Every species contains thousands of molecular solutions in its DNA. Our mission is to build open tools decoding these solutions to lift all boats (pun intended), empowering global communities to find cures like Max hopes for his father.

Thanks to Claude for editing.

I'm Olga Botvinnik, and I'm thrilled to welcome you to Seanome's journey. As the founder of this non-profit focused research organization (FRO), I'm here to share our mission to revolutionize how we understand and learn from ocean biodiversity.

Why the ocean?

The ocean harbors an untapped library of genetic innovation, refined by 4 billions years of evolution. As the primary home to 70% of all animal phyla1, aquatic environments are a living laboratory of innovative body plans, immune systems, and nervous systems far more diverse than on land. Yet today, our understanding of biodiversity remains remarkably narrow — we focus on less than 1% of animal species 2, primarily those directly relevant to drug development. The inequality in our knowledge is staggering: human genetic data is 2,000,000x more enriched in high-quality sequence databases3 — we’re talking Egyptian Pharaoh levels of inequality here — than it would be if our research efforts were distributed equally across species.

Life underwater faces extraordinary evolutionary pressures: marine organisms battle an unprecedented barrage of pathogens, with 70% of marine biomass being microbial4 and a single teaspoon of seawater holds a staggering 100 million viruses5. This has driven the evolution of sophisticated immune systems with potential applications from novel antibiotics to innovative therapies. Marine biology has already revolutionized our field once. Green Fluorescent Protein (GFP), initially isolated from jellyfish in 19626, became biology’s microscopic flashlight in the 1990s7, leading to a Nobel Prize for its discovery. Today, GFP-like fluorescent proteins are are indispensable tools in virtually every cell biology lab worldwide. Like GFP, revolutionary discoveries swim past us every day. A vast molecular library lies beneath the waves — we're building the computational tools to decode its ancient wisdom.

Tools for Biodiversity Breakthroughs

At Seanome, we decode biology's greatest mysteries: genes whose functions remain completely unknown. These "mystery genes" resist classification by commonly used tools like BLAST and HMMer, and even by sophisticated AI approaches such as FoldSeek, returning zero results — no homologs, no structural matches, no interpretable patterns. Where existing methods reach their their detection limits, we're developing sensitive algorithms to reveal hidden molecular functions.

Our mission is to unlock breakthroughs from ocean biodiversity by building open genomics tools

More technically, we build tools to find genes with similar functions but very different sequences (”functional homologs”). For example, the apoptosis regulators BCL2 (human) and ced9 (Caenorhabditis elegans) have only ~20% sequence similarity (much lower than the 50% minimum commonly used for identifying homologous genes!), and yet, they are functionally interchangeable. If BCL2 is deleted in human cells and ced9 is inserted → get the same function, and similarly, if ced9 is deleted in C. elegans, and human BCL2 is inserted→ get the same function8.

Our ultimate goal is ambitious: to decode every protein in the living world. We're starting with the ocean — a deep reservoir of undiscovered molecular innovations. By building open tools to reveal hidden molecular patterns, we're democratizing access to nature's library — accelerating global discovery of nature’s solutions to complex challenges.

Open-sourcing life’s code

At Seanome, we build computational frameworks that amplify scientific discovery. Using Nextflow, Python, and Rust, (and maybe even more languages!) we create open-source tools that empower thousands of scientists worldwide. Each research project we undertake serves dual purposes: advancing specific scientific knowledge while developing tools to empower the broader scientific community to make more discoveries, faster.

Our vision is to open-source the code of life, empowering humanity’s transformation to a world guided by nature's ancient wisdom

We're at a crucial turning point, where biodiversity is becoming not just about conservation, but our first line of defense against challenges ranging from disease, climate, or personal care.

What’s Next: Arctic Clams and Beyond

Our journey begins with an extraordinary discovery driven by personal mission. When Max Glanz, our collaborator at University of Florida, set out to find a cure for his father's rare neurodegenerative disease, he made a remarkable finding: Arctic clams naturally tolerate lipids toxic to human neurons. Next week, we'll dive into the science behind this remarkable adaptation.

Thanks to Claude for editing.

Footnotes

2

The “2,000,000x enriched” number comes from here: As of Nov 14th, 2024, UniProtKB Reviewed (Swiss-Prot) contains 109,973 entries for animals [metazoans, (taxonomy_id:33208)], and 20,428 of those are human. 20,428/109,973 = 0.18575…, which we can round to 0.2. Then, we assume equal distribution and similar numbers of genes for all 10,000,000 predicted animal species (see reference 3 below), which maybe an overestimate, but a useful order of magnitude. If human is 1/10,000,000 animals species, then we calculate enrichment as 0.2/(1/10,000,000) = 2,000,000.

UniProt KB Search Screenshot: