Everything your team needs to prepare: a hands-on guide to computational biology tools, the structure of the final presentation, the toolbox and all the frequently asked questions.
A practical walkthrough for teams using computational biology tools for the first time, from protein sequence to structure prediction and evaluation: what to do at each step, which tools to use and how to judge how reliable the results are.
Start from a clear, testable biological question — not from the tool. Decide which protein (or complex) you are interested in, what you will compare it with and what you will look at to answer.
A good fit for structure prediction
Needs other methods too
Look the protein up in UniProt and download its amino-acid sequence in FASTA format. Check the entry’s annotation — organism, isoforms and features such as a signal peptide. For the variant, change only that one residue in a copy of the sequence.
Caution: A residue’s position in the literature can differ from its position in the UniProt sequence — β‑lactamases are commonly described with Ambler numbering. Before substituting, make sure the position you picked really holds the expected amino acid.
Before running any prediction, see whether structures already exist. The AlphaFold Protein Structure Database holds ready-made predictions for many UniProt proteins, while RCSB PDB and PDBe hold experimentally determined structures.
AlphaFold Server gives free access to AlphaFold 3 for non-commercial use, with a daily job limit. You paste the sequence (or the sequences of the complex), name the job and submit it. Run the reference and the variant as two separate jobs so that they stay comparable.
Caution: AlphaFold Server accepts only the ligands, ions and modifications it supports — not arbitrary chemical structures. Check its FAQ before you design your workflow; complexes with inhibitors such as avibactam may need a local AlphaFold 3 installation or docking.
Every prediction comes with confidence scores. They are the first thing you look at: they show which parts of the model you can use and which you cannot. A structure with low confidence exactly where your region of interest lies does not support conclusions about that region.
Open the models in ChimeraX or PyMOL, colour them by pLDDT and superimpose them on the experimental structures. The RMSD after superposition gives a numerical measure of the deviation; Foldseek finds structurally similar proteins, even where sequence similarity is low.
Useful commands
color bfactor #1 palette alphafoldColour by pLDDT (the values are stored in the B-factor field).matchmaker #2 to #1Superimpose model #2 on #1; the RMSD is reported.align variant, referenceSuperimpose two structures and compute the RMSD (object names are illustrative).AlphaFold gives a static structure prediction. It was not designed to predict the effect of point mutations — a single substitution often barely changes the predicted structure — nor binding affinities. That is not a failure of your team; it is the limit of the method, and it should be stated.
A suggested structure for your team’s final presentation: four parts that take the audience from the idea to the results.
about 1 min
about 3 min
about 3 min
about 3 min
A rule for every team
In the final presentation, every team must report:
Using AI and computational biology tools does not replace scientific validation.
A curated list of bioinformatics and computational biology tools, databases and resources: from checking your sequence to predicting, comparing and visualising structures.
Start here: the tools that cover the core of the hackathon workflow.
Official web access to AlphaFold 3 for non-commercial use.
alphafoldserver.comCheck protein sequences, isoforms and functional annotation.
uniprot.orgSearch available predicted protein structures and complexes.
alphafold.ebi.ac.ukExperimentally determined 3D structures and structural data.
rcsb.org3D visualisation, analysis and comparison of molecular structures.
rbvi.ucsf.edu/chimeraxShowing all 12 tools
Stage 1 · 4 tools
Check and align sequences, and find homologous proteins and functional regions.
Homology search and local sequence similarity.
Protein families and functional regions.
Multiple sequence alignment.
A collection of tools for sequences, alignment and analyses.
Stage 2 · 4 tools
Predict the structure of proteins and complexes from sequence.
Official AlphaFold 3 code and technical documentation; the terms of use for the model parameters apply.
An accessible research workflow for predictions based on AlphaFold and derived methods.
An alternative tool and model for predicting protein structures and interactions.
Homology modelling of protein structures.
Stage 3 · 1 tool
Look up predicted and experimentally determined structures.
The European portal for Protein Data Bank data and structural biology tools.
Stage 4 · 1 tool
Find proteins with a similar three-dimensional structure.
Fast structural similarity search.
Stage 5 · 1 tool
Inspect, analyse and present structures in three dimensions.
Molecular visualisation and creation of scientific figures.
Reference material · 1 resource
Official documentation and resources for the structure prediction tools.
Main page, documentation and access to AlphaFold resources.
The list is indicative and extensible; the Scientific Committee may add or remove tools according to the needs of the four challenges.
Documenting your tools in the final presentation: see the rule for every team
These frequently asked questions explain how the Hackathon is run, the basic requirements and the stages of the competition, so that you know what to expect and can prepare properly for taking part.
31 questions
Suggested searches
BIOHackathon 2027 is a computational biology hackathon organised jointly by the following institutions: Region of Central Macedonia (Deputy Region of Digital Governance), WALK – Center for Entrepreneurship and Innovation of the Aristotle University of Thessaloniki, International Hellenic University (Department of Nutritional Sciences and Dietetics), International Hellenic University (Department of Biomedical Sciences), Centre for Research and Technology Hellas (Institute of Applied Biosciences, INAB), National Hellenic Research Foundation (Institute of Chemical Biology), University of West Attica (Department of Biomedical Engineering), Anatolia American University U.L.E., ACM Student Chapter UoA and Google Developer Group on Campus Athens.
Its purpose is collaboration, developing ideas and searching for innovative solutions to specific challenges (cases). Working in teams, participants use computational biology tools on four antimicrobial-resistance challenges: health, food, agriculture and livestock.
No. Participation is free of charge for all participants.
You register individually, but you always take part as a member of a team. Every participant fills in the organisers’ participant registration form themselves, with their own details, and accepts the terms of participation individually.
Solutions are developed and presented only by teams of 2 to 5 members; there are no individual solutions. If you don’t have a team, simply register and the Organising Committee will place you in one.
No. You can register without a team: the Organising Committee will place you in one, with the aim of interdisciplinary collaboration. Everyone takes part as a team member, though, because solutions are developed and presented by teams only.
Each participant takes part only in the team in which they have registered or been placed.
The hackathon welcomes students, researchers and professionals interested in biology, artificial intelligence and data analysis.
Who is eligible: Natural persons aged 18 or over: students, researchers, members of the academic community, scientists, professionals and anyone else interested (Terms, Article 1).
Each participant registers individually, accepts the terms of participation and becomes a member of a team of 2 to 5 people from different scientific or professional fields — for example, biology or biochemistry knowledge combined with computational skills.
Teams are interdisciplinary, so their members complement each other. The programme includes talks by experts that provide the necessary scientific and technical background, and technical mentors help with preparing input data, reading the outputs and assessing predictions. The preparation resources also include a guide to computational biology tools and a toolbox.
If you have never worked with structure prediction tools, it is worth trying a first prediction — for example on AlphaFold Server — before the hackathon.
BIOHackathon combines short talks with hands-on teamwork on specific cases and real challenges. Teams use computational tools to explore the cases, develop their own approach and propose well-founded solutions.
Key stages:
Each team’s deliverables and their deadlines:
A rule for everyone: state the main tools, data sources, libraries, models and third-party material you used, together with the key inputs and parameters.
Final presentation — suggested guide:
Suggested slide structure:
The winners are chosen by the Jury of BIOHackathon, based on the evaluation criteria. The evaluation takes place at the main event: 26 February 2027.
The members of the Jury are listed on the Committees and mentors page.
Yes, provided you have the right to use it for the purpose of your participation. Under the terms of participation:
For example, AlphaFold Server permits non-commercial use only, under its terms of service (opens in a new tab).
Yes. Participants follow the principles of scientific ethics, transparency, mutual respect and good-faith collaboration. Harassment, discrimination, intimidation and obstructing other people’s participation are not allowed. Knowledge sharing and collaboration between teams are encouraged, but must not lead to unauthorised copying or appropriation of other people’s work.
Grounds for exclusion include, for example, a material breach of the terms, plagiarism, submitting false or misleading information, deliberately falsifying data or results, unauthorised use of software, data or tools, a serious breach of the rules of ethics or safety, and any attempt to unduly influence the evaluation. Before an exclusion decision, the person concerned may set out their views in writing within 24 hours of being notified of the breach.
The Jury assesses the proposals and solutions against 4 criteria, with the following weights:
Each criterion is scored from 0 to 10; the final score is the weighted sum. In the event of a tie, the team with the higher score in criterion A ranks first; if the tie remains, criterion D decides.
The aim of the evaluation is to highlight solutions with real impact, an innovative approach, scientific validity, a direct link to the challenge and an effective presentation.
Yes. The following cash prizes are awarded to the winning teams:
Each prize is paid in equal shares to the members of the winning team, into each member’s own bank account. Payment follows the applicable law and is subject to any taxes or statutory deductions; the necessary supporting documents may also be requested.
Award ceremony: 26 February 2027
Those who create it — you and your team. Participants keep any intellectual or industrial property rights they hold in the original works, methodologies, code, analyses, presentations and other results they create within the Hackathon. Taking part does not transfer those rights to the organisers, the mentors or other participants; mentors, in particular, waive any intellectual property rights in the solutions of the teams they guide.
You grant the organisers a non-exclusive, royalty-free, worldwide and perpetual licence to use, present, publish and reproduce the deliverables, presentations and summaries of the solutions, for promotion, information and education. The licence does not include any right to exploit them commercially or to publish your full source code, confidential material or unpublished research data.
Presenting an idea or solution at the Hackathon does not give other participants any right to copy or appropriate it.
Presenting a solution in public may affect whether some results can be protected later (e.g. by a patent). If this concerns you, take the steps you need in good time.
Before the main event, teams are formed, meet their mentors in the Kick Start meetings and submit their proposals, which the Jury evaluates — see the key stages.
By filling in the organisers’ participant registration form (opens in a new tab). Registration is individual: every participant registers with their own details and accepts the terms of participation individually.
Either way, the solution is developed and presented by the team. Mentors register separately, using the mentor registration form (opens in a new tab).
Registration is open. Registration deadline: 30 November 2026.
Yes. Mentors register separately, as scientific or technical mentors, using the mentor registration form (opens in a new tab). They guide the teams as each team shapes and develops its approach.
For any questions about the event, contact the Organising Committee:
During the hackathon, scientific and technical questions go to your team’s mentors.
Each team has 2 to 5 members from different scientific or professional fields, to encourage interdisciplinary collaboration. Every member registers individually, giving the team’s name; people who register without a team are placed in one by the Organising Committee. Each team appoints a representative, who is its main point of contact with the organisers.
One. Each team chooses one of the four antimicrobial-resistance axes:
Each axis comes with a short description and an indicative research question. The questions are not mandatory: each team may frame its own, as long as it is clearly linked to the challenge it has chosen.
Each team is supported by one scientific and one technical mentor. The Organising Committee does the matching, based on the teams’ needs, the relevance of each mentor’s expertise and their availability, taking any preferences into account.
Mentoring takes place online, according to the availability each mentor states. Mentors only advise: responsibility for the methodology, the results and the submission of the solution stays with the team.
Every participant submits only their own details, through the organisers’ registration form; the team name is used only to group members of the same team. How your information is processed is described in the privacy notice, together with your rights under the General Data Protection Regulation (GDPR).
AlphaFold predicts the 3D structure of a protein from its amino-acid sequence and gives every prediction confidence scores for each part of the structure. AlphaFold 3 can also model complexes of proteins with other proteins, nucleic acids, small molecules and ions.
What it does not do:
The indicative research questions of the challenges often involve mutations (e.g. in a carbapenemase such as KPC, or in the CYP51 protein). Differences between the reference protein and a variant can be subtle, so combine predictions with experimental structures and complementary tools.
FASTA format from UniProt.AlphaFold Server supports a limited set of ligands and ions; for other small molecules, a local AlphaFold 3 installation accepts definitions as a CCD code or SMILES string.
Watch the numbering: mutation positions in β-lactamases are usually given in Ambler numbering, which may differ from the position in your own sequence.
They are the scores AlphaFold uses to estimate how reliable its own prediction is.
pLDDT0–100 · per residue (per atom in AlphaFold 3)
Local confidence for each part of the structure, usually shown on models with these colours.
PAEÅ · per residue pairpTM0–1 · whole structureipTM0–1 · interfaceHigh confidence does not necessarily mean that the prediction reflects the biologically relevant state of the protein.
It depends on the stage of your analysis. For example, from the hackathon toolbox:
In your final presentation, cite every tool and database you used.
Biology, chemistry, biochemistry, medicine, bioinformatics, computational biology
Hands-on experience with the computational tools
Some mentors have dual expertise and cover both areas. When an issue spans both, tell your mentor what you are trying to answer and what you have already tried.
No. Using AI and computational biology tools such as AlphaFold does not replace scientific validation. Predictions are working hypotheses, to be checked against experimental data, comparisons and the literature.
In the final presentation, every team documents the tools, databases and sources it used, together with the key inputs and parameters, and presents how it validated its approach and what its limitations are.
No matching questions
Try a different or broader word — or send us your question directly.