We're offering free 1:1 SeqHub training sessions, tailored to your research. New to the platform or already using it, book a time and we'll walk through what's most relevant to your work.
calendar.app.google/KLFsKvszMMko...
Add collaborators to share data you've curated in SeqHub!
Invited collaborators can...
- View functional annotations
- Add their own annotations and experimental results
- Work with the SeqHub Agent
- Run FlashPPI against the data
No longer need to download and send files back and forth.
We'll be at the MIT Microbiome Symposium this Friday, April 17!
If you're interested in learning more about our research at Tatta Bio or want to see how SeqHub can support your work, come find us at our table or set up time to chat (DM or email us at team@tatta.bio).
Collaboration is now live on SeqHub.
🔹 Add labmates to your datasets so you can build upon the same annotations and data with additional research findings.
🔹 Add co-authors to public datasets to credit everyone who contributed to the work.
Registration is open for our April 15 webinar on protein-protein interaction prediction with FlashPPI.
The session will cover how the model works (high-level) and how you can access it in SeqHub. Experimental and computational folks welcome!
Sign up: forms.gle/iBQrpYnLeiF1...
You can now run a multiple sequence alignment directly on your SeqHub search results.
Once the alignment runs, conserved regions are mapped onto the 3D structure of your query protein. Explore results in SeqHub or download as an a3m file.
Alex will help guide SeqHub's scientific and strategic direction, bringing deep expertise in protein sequence data, community resources, and how foundational infrastructure can best support discovery at scale.
We're thrilled to welcome @alexbateman1.bsky.social to the SeqHub Advisory Board!
Alex has played a central role in building core global resources for protein science, including UniProt. He has been a longtime leader in developing standards and infrastructure that underpin modern protein analysis.
Most protein-protein interaction tools work on protein pairs. FlashPPI runs at proteome scale and now across two proteomes at once.
Upload any two datasets (full genomes, partial genomes, or custom protein sets) and get back a predicted interaction network spanning both.
We're hosting a live walkthrough of FlashPPI in SeqHub on April 15 at 11am EST.
We'll briefly discuss our protein-protein interaction model then walk through how you can use it in SeqHub.
Register here: forms.gle/iBQrpYnLeiF1...
That context makes it easier to understand the RNAs upstream and downstream of a protein of interest, start to assess conservation of non-coding regions across genomic contexts, and get a more complete picture of the regulatory and functional systems around a coding sequence.
SeqHub now surfaces Rfam annotations directly in the genomic neighborhood view: family name, accession, length, and a link out to the Rfam entry.
SeqHub now detects non-coding RNAs in genomic context.
Non-coding RNAs are annotated directly in the genomic neighborhood view, giving a more complete picture of operon structure without having to cross-reference external annotation tools.
Sequence and structural alignment in SeqHub just got easier.
Click any similar sequence from your results, you'll get prompted to run an alignment, and then you'll get identity score, coverage, TM-score, and RMSD.
Shout out to the users who flagged this one!
To run this workflow in SeqHub: upload a genome or set of proteins → run FlashPPI (from tool bar) → select an interaction of interest → run CoSearch on that pair.
FlashPPI predicts protein-protein interactions from sequence. CoSearch finds where those proteins co-occur across genomic contexts.
When co-occurrence holds up, it helps you build confidence in what FlashPPI predicted.
📍 MT Partnering Forum | March 18–19
📍 MGE Computational Workshop | March 18–20
We're in Copenhagen this week for the Microbiome Times Partnering Forum and the MGE Computational Workshop.
Our Chief Scientist @microyunha.bsky.social is speaking at MGE on 3/20 about genomic language modeling, including a look at FlashPPI. Come find us if you'll be there or DM us to meet up!
Every predicted protein-protein interaction in SeqHub comes with a residue-level contact map showing exactly how two proteins interact.
High-probability interactions are highlighted to help you pinpoint precisely where contact may occur.
This is FlashPPI's proteome-wide interaction network in SeqHub. We automatically group predicted protein-protein interactions (PPIs) into functional sub-networks and map them onto an interactive genome browser. Users can continue to explore by clicking into clusters.
Proteome-wide PPI screening usually takes days or months.
With FlashPPI it can run in minutes, which makes workflows like this easier to explore:
• checking interaction partners for hypothetical proteins
• mapping protein complexes across a proteome
• screening environmental & metagenomic genomes
If you haven't tried it yet, you can run proteome-wide PPI prediction on your own genome in SeqHub.
Thanks to Decoding Bio for covering FlashPPI in BioByte this week! Good write-up if you want the context behind the method: open.substack.com/pub/decoding...
Drill into any interaction and you get a residue-level contact map (not just whether two proteins interact, but where).
Try it on your own genome or start with our sample genome first: seqhub.org/tattabio/eco...
Preprint: www.biorxiv.org/content/10.6...
Within minutes of uploading your genome, you have a full predicted interaction network with proteins grouped into functional sub-networks. The genome browser stays integrated, so you can toggle between the network and your genomic data without losing context.
FlashPPI is our new proteome-wide protein-protein interaction prediction model. If you haven't had a chance to try FlashPPI yet, here's what the end-to-end flow looks like in SeqHub:
More, high-level context in our blog: seqhub.org/blog/flashpp...
Full technical details in our preprint: www.biorxiv.org/content/10.6...
We’ve made this freely available for non-commercial use in SeqHub, where results are integrated with functional annotations, genomic context, and more.
Protein interaction prediction doesn't scale to full proteomes. It can take days to months to run all pairwise predictions for a single organism. FlashPPI (our new model) works directly from sequences without pairwise comparisons, which can translate to PPI predictions in minutes.
