Although ETH Zurich takes all possible care to ensure the correctness of published information, no warranty can be accepted regarding the correctness, accuracy, up-to-dateness, reliability, and completeness of the content of this information. ETH Zurich expressly reserves the right to change, delete, or temporarily not publish the contents wholly or partly at any time and without prior notice. Liability claims against ETH Zurich for material or immaterial damages arising from access to, use or non-use of the published information, through misuse of the connection, or as a result of technical failures are excluded.
ETH Zurich has not verified third-party websites — that is, websites not located on its servers or within its area of influence — which may be linked to this website. ETH Zurich therefore accepts no responsibility for the content or services offered thereon, nor for compliance with data protection regulations or the absence of harmful software on such websites.
All online documents and web pages as well as their parts are protected by copyright. They may be copied and printed only for private, scientific, or non-commercial use. Without prior written consent from ETH Zurich, it is not permissible for documents or parts thereof to be copied, stored on other servers, input into newsgroups or online services, or saved on CD-ROMs or other data carriers. This applies in particular to the ETH Zurich logo. Temporary storage in a cache or proxy server to optimise access speed is permitted. Citation of documents and web pages with correct source attribution and linking to this website is expressly allowed.
You may visit this website without providing any personal information. When accessing the site, certain technical data such as your IP address, browser type, date and time of access, and the last page visited are automatically recorded in server log files. This information is collected to ensure the security and stability of our services, to analyse usage in an aggregated and anonymised form, and to continuously improve the website. These data do not allow direct identification of individuals and are not combined with other sources.
We use Google Analytics 4 (Google) in a cookieless configuration: it stores no cookies or other identifiers on your device, Google signals and advertising personalisation are switched off, and access keys contained in page addresses are removed before any data is sent. To load the analytics script and send page views, your browser connects to Google's servers, which necessarily receive your IP address; according to Google, Google Analytics 4 does not log or store IP addresses. The statistics we receive are aggregated and do not identify individual users. No third-party cookies are used on this website.
To detect and fix errors, our servers send technical error reports to Sentry (Functional Software, Inc., United States), which stores them in its data centre in the European Union (Germany). A report describes the error and where it occurred and may, in rare cases, contain parts of a submitted query; request contents, e-mail addresses, IP addresses, access keys and other credentials are removed before it is sent. Error reports are used only to keep the service working and are deleted by Sentry after at most 90 days.
To protect the feedback form against automated abuse, sending feedback runs a bot check, Cloudflare Turnstile, provided by Cloudflare, Inc. (United States). It runs in your browser and processes technical information such as your IP address and characteristics of your browser and device to distinguish people from automated traffic. It is loaded only when you send the form, not when you merely visit the page, and the content of your feedback is not sent to Cloudflare. The AI chat assistant uses the same check (see below).
For the operation of the database search service, certain data must be stored temporarily. When a user submits a query, the text of the query is stored in plain text, and the associated e-mail address, if provided, is stored in encrypted form and used only to send status updates about that search. This information is retained to ensure the functionality of the service and to allow users to manage or delete their own query records.
Search records normally expire after 48 hours. Locking a search guarantees retention for at least six months; it may remain available afterwards, but may be deleted at any later point. Re-locking renews the guarantee. You can unlock and delete your search earlier through the user interface. Anyone who has the link to a search can view its results — including the search name, the names of the submitted sequences (from their FASTA headers) and any AI-generated summary — whether or not the search is locked; locking only extends how long the search is kept. The query sequence itself is shown only to the person who submitted the search, or to whoever they give a link that explicitly includes the sequence. When a record expires or is deleted, the corresponding query string and e-mail address are removed from our systems. If the same e-mail address is associated with other query records still within their retention period, those will be retained until their respective deletion conditions are met.
In addition, we keep pseudonymised usage statistics for every search: one-way hashes of the e-mail address, IP address, browser identifier and query, together with technical information such as the databases searched, the size of the query, run times, and the use and cost of AI features. They contain neither the query nor the e-mail address in readable form, are not linked to your identity, and remain after the search itself has been deleted. They are used only to provide the service — for example to monitor its load and capacity and to detect abuse — and are never used for marketing or shared with third parties.
In addition, the website keeps some information in your browser's local storage — on your device, not on our servers, and without using cookies — so that it keeps working across visits:
This information is not shared with third parties. You can remove it at any time by clearing this website's data in your browser settings. Anyone with access to your browser profile may be able to see it.
All data are processed under the responsibility of ETH Zurich. The search indexes run on ETH Zurich servers in Switzerland. The search service and its databases are hosted by Railway (Railway Corporation, United States) in its data centre in the European Union (Netherlands). The complete result sets of large searches are stored with Amazon Web Services (Amazon Web Services, Inc., United States) in a data centre in the United States; they contain search results, not your query or e-mail address. Status e-mails are sent through ETH Zurich's Microsoft 365 service. Appropriate technical and organisational measures are in place to protect data from unauthorised access, misuse, loss, or disclosure. Once a record has been deleted, it cannot be recovered.
When you run a search, you may optionally request an AI-generated summary of the results and/or an AI-assisted organism identification. These features are opt-in and are not used unless you enable them for a given search.
To produce them, information about your search results is transmitted to and processed by a third-party artificial-intelligence provider (currently OpenAI, United States, using GPT models). The information sent consists of derived metadata about the matched public samples — such as organism names, sample accessions, geographic and project information, and dates, obtained from public sequence archives — together with any text contained in the descriptive headers of the FASTA sequence you submitted. Your raw query sequence itself is not sent, and no e-mail address, IP address, or account identifier is included. This processing involves transferring data to a provider located outside Switzerland and the European Economic Area; by enabling these features you consent to this transfer.
The generated summary and organism identification, the text exchanged with the AI provider, and any feedback you give on a summary (a rating and an optional comment) are stored together with your search and follow the same retention rules described above. You can unlock and delete the search earlier through the user interface.
AI-generated content may be inaccurate or incomplete and should be verified against the underlying results. Because the descriptive header of a submitted sequence is transmitted to the AI provider, please do not include identifiable, personal, or confidential information in FASTA headers.
This section applies specifically to the optional AI chat assistant offered on this website as an experimental beta feature, and supplements the sections above. Using the chat is voluntary — the search service can be used without it. Where the terms below differ from the general policy (in particular regarding processing by a third-party AI provider and the retention of conversations you rate or share), the terms in this section govern the data you provide through the chat.
When you use the chat, the content of your messages and any sequences you upload or paste are processed to generate a response. To do so, this content is transmitted to and processed by a third-party artificial-intelligence provider. Which provider depends on the model selected for the conversation, which is shown in the chat interface before the conversation starts:
To answer, the assistant may look up public databases such as NCBI, ENA, UniProt and Ensembl, sending them search terms or accession numbers taken from your question.
To protect the chat against automated abuse, starting a conversation (or creating a share link) runs a bot check, Cloudflare Turnstile, provided by Cloudflare, Inc. (United States). It runs in your browser and processes technical information such as your IP address and characteristics of your browser and device to distinguish people from automated traffic. It is loaded only at that moment, not when you merely visit the page. We do not send the content of your messages to Cloudflare.
The chat service itself is operated on third-party cloud infrastructure (Railway, United States, in its data centre in the European Union). Using the chat therefore involves transferring your input to service providers located outside Switzerland and the European Economic Area, and by using it you consent to this processing and transfer.
The chat service keeps technical logs — for example timings, costs, errors and your IP address, but not the text of your messages or of the replies — to operate the service, to monitor its costs, and to protect it against misuse. These logs are kept for up to 30 days and then deleted. Please do not enter identifiable human or patient sequences, personal data, or confidential information into the chat, as it will be processed by a third-party provider named above.
We retain chat operational metadata for 30 days, with scheduled cleanup: model and effort, token and estimated cost totals, timing, success or failure, tool names and counts, linked search identifiers, and a pseudonymised (keyed-hash) form of your IP address. This operational record does not contain ordinary message text, answers, or sequence contents. It is available to authorized administrators to monitor reliability, costs, and abuse.
For cost accounting we also keep a record of each AI model call without identifiers — its time, model, token counts and cost — for up to 24 months, and daily totals indefinitely.
If you rate a reply (thumbs up or down, optionally with a written comment), your rating and any comment, together with the associated conversation — including any sequences it contains — may be stored and used to evaluate and improve the quality of the service, including the prompts and instructions it uses. Such data is retained for up to six months and is then deleted. This use is based on your consent, which you may withdraw at any time by contacting us at the address below.
If you choose to create a share link for a conversation, a snapshot of that conversation — including any sequences it contains — is stored and made accessible to anyone who has the link for a limited period (currently up to two days), after which it is automatically deleted.
The chat is an experimental beta feature. The providers involved, its behaviour, and the retention periods described here may change. Its responses may be inaccurate and should always be verified against the linked search results.
The processing of personal data described above is carried out in accordance with the Swiss Federal Act on Data Protection (revDSG) and, where applicable, the EU General Data Protection Regulation (GDPR). The legal bases for processing are the user's consent (Art. 6 para. 1 lit. a GDPR) and ETH Zurich's legitimate interest in providing and maintaining a secure, reliable, and efficient online service (Art. 6 para. 1 lit. f GDPR). Users may withdraw their consent at any time by deleting their data or contacting ETH Zurich.
Under applicable data protection law, you have the right to access, rectify, and delete your personal data, as well as the right to restrict or object to its processing. You also have the right to data portability in a commonly used, machine-readable format, where applicable. Requests relating to these rights should be directed to the contact address below.
If you believe that your data is being processed in violation of data protection law, you may lodge a complaint with the Federal Data Protection and Information Commissioner (FDPIC) or, if you reside in the EU, with the competent supervisory authority in your member state.
For questions regarding data protection or this privacy policy, please contact:
ETH Zurich
Data Protection Officer
Rämistrasse 101
8092 Zurich, Switzerland
E-mail: datenschutz@ethz.ch
The Biomedical Informatics Group has no influence over external sites to which this website links. Therefore, with regard to such sites, it can neither guarantee that the information contained therein is correct nor that the content is free from malware (e.g., viruses) nor that data protection legislation is observed there. Nor can it take any responsibility for ensuring that these links always function correctly or continue to refer to the same content. The Biomedical Informatics Group likewise has no control over sites that link to its services and therefore cannot accept any responsibility for them.