Strategic Plan 2025-2027

Organisation overview

Mission and vision

The Ersilia Open Source Initiative is a Catalan non-profit organisation working to eradicate infectious and neglected diseases, often termed “diseases of poverty”. Our mission is to fuel sustainable research across the Global South, where those diseases are more prevalent, through the development and application of novel artificial intelligence (AI) and data science tools.

We envision a world with a thriving and equitable scientific research ecosystem where no barriers, tangible (lack of infrastructure, insufficient funding, etc.…) or intangible (inaccessibility to knowledge, isolated environments, neocolonialist practices, etc.), hamper the development of new treatments for diseases that affect the most underprivileged sectors of society.

Key priorities

To develop cutting-edge scientific activity in research areas that are severely underfunded and neglected by the pharmaceutical industry and the Global North research agendas.
To bridge the knowledge gap in AI and data science for researchers working in low and middle income countries (LMIC).
To increase the scientific productivity and agency of researchers working locally in the Global South investigating endemic diseases.
To promote open science and open source research software as a means to truly reduce the existing imbalance in global health.

To learn more about how we have defined our key priorities, please read Ersilia’s Ten Principles.

Core values

Openness. Scientific knowledge is a public good and should remain open, free and accessible to all. Privatised science only benefits a privileged minority and stalls innovation, affecting especially research areas with poor economic incentives. Working radically in the open is one of Ersilia's core tenets.

Innovation. Creative thinking is at the root of scientific discovery. To truly advance in the understanding and cure of infectious diseases, bold ideas and hypotheses must be formulated and tested. Cutting-edge science can happen in, and for, the Global South.

Collaboration. Global health is a vast research field. It is beyond the capacity of Ersilia to tackle it in isolation. We advance in our mission thanks to our collaborative spirit and community building efforts. By establishing key partnerships with different actors (academic researchers both in the Global North and the Global South, other research foundations, governmental agencies and pharmaceutical companies), we multiply our effectiveness and maximise the impact of our programs.

Sustainability. For long, science has been rooted in colonialist and extractivist practices, which have prevented the full development of research institutions in the Global South. By prioritizing capacity building, training and knowledge transfer we ensure that Ersilia’s impact will not be limited to a single project or program, but will set the foundation for growth and scaling at partner organisations.

Integrity and accountability. Implicit in the work of any scientist is the trust in the scientific method and the critical analysis of results. We conduct research with the highest standards of integrity and take responsibility for our findings.

Needs assessment

Global Health broadly encompasses all work, research and efforts to improve health and achieve health equity for all people worldwide. However, health equity is still far from real. Communities in low income countries (LIC) suffer from a disproportionate burden of infectious diseases, with six of the top ten causes of death in LIC being still due to infections. Diseases mostly affecting low-resourced regions are less researched, as returns on investment for pharmaceutical companies are not sufficiently incentivising. Indeed, only 15% of the drugs in development worldwide target infectious diseases, which include malaria, tuberculosis, COVID, and HIV, but also a long list of neglected tropical diseases (Global Health Observatory, World Health Organization (WHO)). While most scientists in those regions strive to find new cures for endemic diseases, historical inequalities have placed them at a severe disadvantage, with their projects globally amounting to less than 5% of the world’s research (assessed by the number of publications, NSF, 2023).

As the WHO put it, “All countries therefore need to be producers of research as well as consumers of it.” (World Health Report, 2013). Ersilia focuses on this particular aspect of global health, empowering researchers across the Global South with AI tools for infectious disease research. As an agile and technology-oriented research organisation we are uniquely positioned to truly strengthen the research ecosystem in low-resource settings, alongside African partners. Ersilia will be established as a nexus for in-country, AI-driven drug discovery research across the Global South. We will achieve this by (a) strictly following the principles of open source, (b) growing local AI hubs at established research centres, and (c) training the next generation of world-class scientists in the Global South. Ultimately, we aim to improve the lives of millions of people by eliminating diseases associated with poverty.

Landscape

Internal strengths:

Development of state-of-the-art computational biology and chemistry tools by incorporating the latest AI developments into our projects.
Translation of research done in well-resourced environments for well-studied diseases into tools for underfunded research in neglected diseases.
Partnerships with local and international research institutions, especially with global health stakeholders.

Internal challenges:

Building a sustainable organisation while remaining true to our free and open source principles.
Balance our openness requirements with the reality of the current drug discovery industry which heavily relies on intellectual property protection.
Attracting and retaining talent in a competitive AI landscape by providing industry-level compensation and career growth opportunities.

External opportunities:

Global commitment to eradicate malaria, tuberculosis and HIV by 2030.
Establishment of research software as a branch of research with unique needs and capacities.
Growing awareness of the need for open science and open source research, and inclusion of open policies by large research funding bodies.

External threats:

Global lack of funding for infectious disease research.
Low data available for most neglected diseases as a result of decades of insufficient research and funding.
Scepticism and misunderstanding of the added value of implementing AI in research due to anticipated hype in the industry.

Achievements so far:

Ersilia was founded in November 2020 by Miquel, Gemma and Edoardo. In four years, we have achieved remarkable milestones in our path to build a sustainable tech non-profit organisation making a measurable impact in global health:

Development of a free, open source resource for AI research in infectious diseases. The Ersilia Model Hub is currently the largest open source repository of ready-to-use AI models for drug discovery in infectious diseases. It contains over 180 artificial intelligence and machine learning (AI/ML) models for several disease areas (malaria, tuberculosis, HIV, COVID-19, schistosomiasis and antimicrobial resistance, among others) and drug discovery-related properties (ADME, toxicity…). Since 2022, new generative frameworks for chemistry have also been added to the Hub. The Ersilia Model Hub is the backend of various tools developed by Ersilia, including ZairaChem (Turon et al, Nat Commun, 2023) and ChemSampler (Turon et al, ACS Med Chem Lett, 2024). The Ersilia Model Hub was recognised by the GitHub for the Social Good Award in 2023 and as a Digital Public Good in 2024.
Training scientists across the Global South and establishing local data units. Adoption of AI in low-resource settings requires skills development opportunities for local researchers. In these four years, we have directly trained more than 100 researchers (chemists, biologists, pharmacists and bioinformaticians) on the applications of AI to drug discovery via in-person workshops (South Africa, Zambia, Cameroon and Ghana) and online mentoring. We are also establishing local data science units with Ersilia’s tools at the H3D Centre (University of Cape Town, South Africa), the Centre for Drug Discovery (UB-CeDD, University of Buea, Cameroon) and the IMPM (Yaoundé, Cameroon). In addition to training scientists, we are also devoted to supporting the next generation of software engineers and data scientists from underrepresented populations. Through participation in the Outreachy internship program and others, we have offered paid internships to 20 students from Africa, Middle East and South America.
Advancing the scientific knowledge on infectious and neglected diseases. We have engaged in several research projects in collaboration with LMIC organisations and mission-aligned institutions. Notably, we have developed and implemented the first AI/ML screening cascade for malaria and tuberculosis drug discovery in Africa, in collaboration with the H3D Centre (Turon et al, Nat Commun, 2023), screened African Natural products for antiviral activity (Betow et al, Mol Inf, 2024), identified new hit candidates with excellent in vitro activity for malaria (Turon et al, ACS Med Chem Lett, 2024), and advanced pharmacogenomics research for African populations (Turon et al, Nature, 2024; Turon et al, medRxiv, 2024). Overall, we have published 23 papers (13 research articles, 4 reviews and 6 commentaries and opinion articles).
Establishment of a network of funders, volunteers and collaborators. Tackling Global Health research questions cannot be done in isolation. We have therefore strived to build a network of supporters and beneficiaries to ensure we stay on track with our mission and vision. On one hand, we have participated in several programs to quick-start our organisation (Fast Forward Accelerator, 2022), and learn more about open source and open science (Digital Infrastructure Incubator, 2021, Open Life Sciences, 2023, Software Sustainability Fellowship, 2023) and engage with tech experts (Mozilla Builders Accelerator, 2024). On the other hand, we have established a solid network of collaborators in the African continent (H3D Centre, Stellenbosch University, University of Pretoria, University of Buea, University of Yaoundé I, IMPM, KEMRI, CIDRZ, The Gambia MRC) and abroad (University of Goias, University of Pisa, The Wistar Institute, CeMM, IRB Barcelona, EMBL-EBI) to perform high impact research. We also highly value the contributions of the more than 50 volunteers who have dedicated their time and expertise to Ersilia in the last four years, independently or via programs from Harvard University (Tech4SocialGood), UC Berkeley, the Good Data Institute, GitHub, Atlassian, Splunk and Digital Ocean.

All those achievements would not have been possible without the generous support of our funders. Collectively, Ersilia has received, either via the UK or the Spanish organisation, funds from the following organisations: Fast Forward, BlackRock, HPE, Splunk, GitHub, Okta, Digital Lift, Roddenberry Foundation, Oakdale Trust, Astor Foundation, FundOSS, The Fore, Code for Science and Society, Rosetrees Trust, Rotary Club, Fundació la Caixa, FAIR-Impact EU, Mozilla Foundation, Merck KGaA, Ministerio de Ciencia, Innovación y Universidades (Gobierno de España). In addition, in partnership, Ersilia has received funds from the Bill and Melinda Gates Foundation (Calestous Juma Fellowship awarded to Prof. Ntie-Kang), AI2050 Schmidt Sciences (Senior Fellowship awarded to Prof. Kelly Chibale), GSK, Novartis and South African MRC (GRADIENT project with the H3D Centre) and Wellcome Trust (Event Fund from CS&S in collaboration with the H3D Centre).

A note on our transition to Spain: Ersilia was originally founded as a Charity in Cambridge, United Kingdom, the place of residence of one of the co-founders. In 2023, Ersilia founded a non-profit organisation in Girona, Spain, under the legal name “Fundació Ersilia Open Source Initiative”. In 2024, the transition from the UK to Spain was completed and the UK Charity was closed. All remaining funds were transferred to the Spanish organisation, now the sole organisation in charge of carrying forward Ersilia’s mission.

Strategic Priorities

Priority 1: To build the reference resource for AI/ML research in infectious diseases

While AI has revolutionised biomedical research, with tech giants like Google’s DeepMind or unicorn start-ups like ExScientia leading the way, it is crucial that this technology remains open for the global majority, and serves their needs. The Ersilia Model Hub is the Foundation’s main asset. It is the first and only open-source repository of AI/ML models specifically designed for biomedical research in global health and with a focus on the discovery of drugs for infectious diseases. We aim to increase the number of models available through the Hub and improve their accessibility.

Goals:

Increase the number of AI models available through the Hub. Currently, Ersilia offers over 180 AI/ML models developed by third parties (75%) and by Ersilia (25%), related to bioactivity prediction, anticipation of toxicity, chemical properties, molecular descriptors, chemical space sampling and generative AI. Broadly speaking, Ersilia contains tools to annotate drug candidates with predicted properties and activities, explore the diversity and scope of chemical libraries, and sample the chemical space in search of new chemical matter. Our focus is to have a comprehensive set of activity prediction models and generative AI frameworks for chemistry, all related to anti-infectives drug discovery.
Improve accessibility and documentation of the Ersilia Model Hub. The Ersilia Model Hub is offered through a downloadable Command Line Interface (CLI) and as a Python Package (PyPi). Users can learn more about how to use Ersilia in our documentation (Ersilia Book). We aim to (a) consolidate the CLI in a release associated with a scientific publication, (b) clearly document all of Ersilia’s functions both for users and developers, and (c) continue the development of an LLM-based Ersilia assistant to facilitate user interaction with the platform.
Provide an online version of the Ersilia Model Hub. To bypass low computational capacity at partner organisations and facilitate its usage, we aim to deploy a selection of AI/ML models for online inference through the AWS cloud.
Complete an effort to pre-compute and store frequent user queries. We aim to save computational resources, provide faster service and enable large scale studies of the chemical space by persistently storing calculations in a fully open fashion.

KPIs:

Number and provenance of models available through the Hub.
Number of scientific publications citing the Hub and demonstrating its usage.
Number of users.
Number of contributors, understood as developers (Ersilia employees, interns, volunteers and open source contributors) as well as model donors.
Number of datapoints stored in our database.

Resources needed:

Personnel. The maintenance and extension of the Ersilia Model Hub requires two software engineers dedicated full-time to this task, and we estimate to engage 5-10 interns per year and an equal number of open source contributors. To expand our community engagement activities, Ersilia would require a project manager or community builder responsible for communicating with, maintaining and organising our open source community.
Infrastructure. Ersilia relies on free and open solutions for the majority of our infrastructural needs. The Ersilia Model Hub is hosted in a GitHub repository and released via PyPi and Conda. We benefit from a pro-bono plan to run GitHub Actions workflows that ensure continuous testing of the platform. In addition, Splunk has generously contributed a Splunk Enterprise license that allows us to monitor model usage and produce monthly reports. For cloud deployment solutions, we currently benefit from pro-bono credits from Digital Ocean, and we are building an AWS infrastructure. The cost of the AWS infrastructure will be covered by grants and pro-bono AWS credits.
Funding. Open source software has been traditionally difficult to maintain. To achieve our goals, Ersilia must raise funding to support the hiring of the detailed personnel as well as cover the costs of the backend infrastructure.
Data. AI model building is data intensive and most drug discovery-related data is locked in IP-protected databases. We liaise with organisations willing to share their data so we can build new models, as well as establish pipelines for automatically curating open data available in repositories and publications. Examples of such organisations would be the Swiss TPH, MMV and the Seattle Children’s Hospital.

Priority 2: To focus on novel therapeutic opportunities and targets for understudied diseases

Recent advances in AI for biomedicine, exemplified by AlphaFold, have erupted into all areas of scientific research, including drug discovery. Two prospects are particularly relevant to Ersilia’s mission. First, AI-enabled structural biology research, which may be particularly transformative in infectious diseases given that, for many pathogens, experimental structural information is scarce or practically non-existent. AI-predicted structures can enable drug discovery research in neglected disease areas. Second, and importantly, we are determined to translate cutting-edge methodology developed in well-funded disease areas such as cancer into antimicrobial drug discovery. This includes new therapeutic modalities, especially targeted protein degradation, as well as omics data analysis. Both are core expertises of the Ersilia team that are sorely lacking in drug discovery for global health. As part of these efforts, both pathogen- and host-directed therapies will be considered.

Goals:

Identify new targets for orphan neglected tropical diseases using an AI-first approach.
Lead the adoption of targeted protein degradation in the field of antimicrobial drug discovery.
Explore the chemical space associated with antimicrobial drug discovery, with a special focus on natural product (-like) compounds.

KPIs:

Number of diseases investigated in our research programs.
Number of validated targets identified.
Number of collaborations with research groups with experimental capacity to test our hypotheses.
Number of scientific publications and associated citations.

Resources needed:

Personnel. Dedicated, highly trained research personnel (PhD students and Postdoctoral researchers) will be required to perform such studies. We will hire at least one new postdoctoral researcher and offer a 4-year PhD position at Ersilia.
Infrastructure. GPU-powered resources (local and cloud) are required to perform these analyses. We will use Ersilia’s own internal resources as well as partner with universities and research centers with access to computer clusters that can fill in our needs.
Funding. Dedicated research funding from governmental and international sources (Spanish Ministry of Science, Innovation and Universities, Horizon EU, Wellcome Trust, NIH and others) will be applied to cover the costs of research and the funding for specific research positions (PhD student stipends, Postdoctoral salaries).

Priority 3: To provide long-term training opportunities for scientists in the Global South

Over the last four years, we have found a greater need than anticipated in the area of capacity building. In practical terms, our beneficiaries (mostly MSc-, PhD-, PostDoc-level researchers) have not received the necessary training to maintain (and further improve) open source AI models, putting them at a clear disadvantage in the global health space, with most of the expertise retained by the Global North. For this reason, we have established two main programs, focused on (1) the creation of local hubs in organisations that can act as amplifiers for their region, and (2) the delivery of capacity building activities adapted to the needs of local researchers. For the latter, we are designing a program to support 6-10 research groups every year in an “AI Incubator” fashion. During the AI Incubator we will accompany early-career researchers in their journey to start adopting AI tools in their ongoing research projects. The Roddenberry Foundation has granted us a Catalyst Award to scope the AI Incubator program, and we are currently running a community consultation to understand user needs.

Goals:

Host at least one in-person workshop each year in a LMIC. Our in-person workshops have a duration of one week and offer an introduction to basic AI/ML concepts as well as hands-on practice on the application of AI models to drug discovery.
Run the first cohort of the AI incubator program. We will host at least 6 research groups working on infectious disease drug discovery and accompany them in their journey to incorporate data science and AI into their research programs. While the first cohort will be focused on drug discovery, we aim to expand to other research areas such as biomarker discovery for future cohorts.
Provide mentorship programs for underrepresented groups in STEM. Introducing the next generation of software developers and data scientists to the realm of open source research software and biomedical sciences allows the growth of local talent that can be incorporated into the working force for science and technology in LMIC. We achieve that via paid internships and mentoring of MSc theses for computer scientists.

KPIs:

Number of scientists trained in AI/ML for biomedical research.
Number of countries targeted by our training programmes.
Number of interns trained.
Number of MSc theses co-directed by Ersilia.

Resources needed:

Personnel. To achieve our training and capacity building objectives we need a combination of skills. On one hand, Ersilia’s researchers must be ready to deliver training, provide online mentorship and engage with the trainees projects, while Ersilia’s software engineers must ensure our technology responds to the users needs and is accessible to them. Finally, administrative personnel (program manager or similar) is key to ensure the success of larger programs like the AI Incubator.
Funding. To achieve our training goals we will rely mostly on philanthropic support from organisations that engage in science for development and open source. To continue our partnership with Outreachy, we need to raise 10.000 USD per each Outreachy intern (8000 USD per salary and 2000 USD to contribute to Outreachy’s overhead costs).

Priority 4: To continue building a community that maximises our impact and ensures our sustainability

We cherish our community and are grateful for all the enthusiasm and support that we have received over the last four years. We aim to continue building our base of supporters, contributors, donors and scientists to achieve our mission. We have also become more aware of the importance of in-person interactions. For this reason, we wish to increase the presence of Ersilia in Barcelona and, by extension, Catalonia, Spain and Europe. This will include the elaboration of outreach materials written in languages other than English, such as Catalan.

Goals:

Establish long-term partnerships with local Global South research institutions to further support outstanding scientists in their home countries who can benefit from Ersilia’s expertise and become stakeholders in the development of data science facilities in their institutions.
Grow a solid network of collaborators and supporters locally in Barcelona. For the first four years, Ersilia has been essentially a remote organization. It is now time to participate in the research ecosystem in our home city, and in Europe more broadly, by engaging in debate, conferences, and activities happening locally.
Build a strong diversity and inclusion work environment. At any given time, between 5 to 20 volunteers and contributors might be working with Ersilia in addition to our full-time staff. These volunteers come from various different countries, backgrounds and expertises, and they are a fundamental part of our work. Ensuring they all feel welcome and safe at Ersilia is paramount to our success.
Expand Ersilia’s advisory board, currently composed of Ersilia’s UK former Trustees, to support our decision-making processes and ensure Ersilia is an organization that answers the real needs of those whom it serves.
Diversify Ersilia’s funding sources. To ensure long term sustainability, we will continue combining our fundraising efforts with grant applications and research funding. We will also expand our donation programs in Spain and the US.
Build internal capacity with key hires in technical and operational roles. To fulfill all the above priorities, increasing Ersilia’s full-time employees with the appropriate skill sets will be required.

KPIs:

Number of partnerships with research institutions, in Africa and elsewhere.
Number of full-time employees at Ersilia.
Number of Advisors and diversity of expertises available.
Funding raised to advance Ersilia’s mission.

Resources needed:

Personnel. The strengthening of Ersilia’s Governance, Community Building and Fundraising will be led by Ersilia’s co-founders Miquel and Gemma with the support of the Board of Trustees.

Financial and resource plan

Ersilia is sustained via a combination of competitive research grants, philanthropic donations, non-profit accelerators and occasional fee-for-service in mission-aligned projects. To achieve our strategic priorities, we have devised an ambitious fundraising plan targeting the different Key Priorities:

Open Source funding. Open source funding for research software from tech organisations (Mozilla Foundation, Splunk, GitHub, Google and others) and governmental initiatives (FAIR-Impact, Horizon EU, OSCARS) will be devoted to achieving Priority 1. The goal is to raise 0.5M USD in the next three years in that area.
Research funding. Scientific programs from the European Union, the United Kingdom, Spain and the United States will be the main target to support the work of Priority 2, along with science philanthropy funders like the Chan Zuckerberg Initiative or the Bill and Melinda Gates Foundation. We aim to secure a large, multi-year grant to expand our scientific program as well as secure funding for academic positions within Ersilia.
Philanthropy. Traditional philanthropy and accelerator-based funding will be devoted to achieving Priorities 3 and 4. The goal is to secure a main funder for the Ersilia AI Incubator by the end of 2025, and to participate in an accelerator program for scale-up stage initiatives by 2026.

Appendix

An up-to-date version of the statistics referred to in this report are available under a CC-BY-4.0 License in the Ersilia Stats repository.

PreviousErsilia privacy notice NextErsilia, the Invisible City

Last updated 1 day ago

Was this helpful?

Strategic Plan 2025-2027

Organisation overview

Mission and vision

Key priorities

To develop cutting-edge scientific activity in research areas that are severely underfunded and neglected by the pharmaceutical industry and the Global North research agendas.
To bridge the knowledge gap in AI and data science for researchers working in low and middle income countries (LMIC).
To increase the scientific productivity and agency of researchers working locally in the Global South investigating endemic diseases.
To promote open science and open source research software as a means to truly reduce the existing imbalance in global health.

To learn more about how we have defined our key priorities, please read Ersilia’s Ten Principles.

Core values

Needs assessment

Landscape

Internal strengths:

Development of state-of-the-art computational biology and chemistry tools by incorporating the latest AI developments into our projects.
Translation of research done in well-resourced environments for well-studied diseases into tools for underfunded research in neglected diseases.
Partnerships with local and international research institutions, especially with global health stakeholders.

Internal challenges:

Building a sustainable organisation while remaining true to our free and open source principles.
Balance our openness requirements with the reality of the current drug discovery industry which heavily relies on intellectual property protection.
Attracting and retaining talent in a competitive AI landscape by providing industry-level compensation and career growth opportunities.

External opportunities:

Global commitment to eradicate malaria, tuberculosis and HIV by 2030.
Establishment of research software as a branch of research with unique needs and capacities.
Growing awareness of the need for open science and open source research, and inclusion of open policies by large research funding bodies.

External threats:

Global lack of funding for infectious disease research.
Low data available for most neglected diseases as a result of decades of insufficient research and funding.
Scepticism and misunderstanding of the added value of implementing AI in research due to anticipated hype in the industry.

Achievements so far:

Development of a free, open source resource for AI research in infectious diseases. The Ersilia Model Hub is currently the largest open source repository of ready-to-use AI models for drug discovery in infectious diseases. It contains over 180 artificial intelligence and machine learning (AI/ML) models for several disease areas (malaria, tuberculosis, HIV, COVID-19, schistosomiasis and antimicrobial resistance, among others) and drug discovery-related properties (ADME, toxicity…). Since 2022, new generative frameworks for chemistry have also been added to the Hub. The Ersilia Model Hub is the backend of various tools developed by Ersilia, including ZairaChem (Turon et al, Nat Commun, 2023) and ChemSampler (Turon et al, ACS Med Chem Lett, 2024). The Ersilia Model Hub was recognised by the GitHub for the Social Good Award in 2023 and as a Digital Public Good in 2024.
Training scientists across the Global South and establishing local data units. Adoption of AI in low-resource settings requires skills development opportunities for local researchers. In these four years, we have directly trained more than 100 researchers (chemists, biologists, pharmacists and bioinformaticians) on the applications of AI to drug discovery via in-person workshops (South Africa, Zambia, Cameroon and Ghana) and online mentoring. We are also establishing local data science units with Ersilia’s tools at the H3D Centre (University of Cape Town, South Africa), the Centre for Drug Discovery (UB-CeDD, University of Buea, Cameroon) and the IMPM (Yaoundé, Cameroon). In addition to training scientists, we are also devoted to supporting the next generation of software engineers and data scientists from underrepresented populations. Through participation in the Outreachy internship program and others, we have offered paid internships to 20 students from Africa, Middle East and South America.
Advancing the scientific knowledge on infectious and neglected diseases. We have engaged in several research projects in collaboration with LMIC organisations and mission-aligned institutions. Notably, we have developed and implemented the first AI/ML screening cascade for malaria and tuberculosis drug discovery in Africa, in collaboration with the H3D Centre (Turon et al, Nat Commun, 2023), screened African Natural products for antiviral activity (Betow et al, Mol Inf, 2024), identified new hit candidates with excellent in vitro activity for malaria (Turon et al, ACS Med Chem Lett, 2024), and advanced pharmacogenomics research for African populations (Turon et al, Nature, 2024; Turon et al, medRxiv, 2024). Overall, we have published 23 papers (13 research articles, 4 reviews and 6 commentaries and opinion articles).
Establishment of a network of funders, volunteers and collaborators. Tackling Global Health research questions cannot be done in isolation. We have therefore strived to build a network of supporters and beneficiaries to ensure we stay on track with our mission and vision. On one hand, we have participated in several programs to quick-start our organisation (Fast Forward Accelerator, 2022), and learn more about open source and open science (Digital Infrastructure Incubator, 2021, Open Life Sciences, 2023, Software Sustainability Fellowship, 2023) and engage with tech experts (Mozilla Builders Accelerator, 2024). On the other hand, we have established a solid network of collaborators in the African continent (H3D Centre, Stellenbosch University, University of Pretoria, University of Buea, University of Yaoundé I, IMPM, KEMRI, CIDRZ, The Gambia MRC) and abroad (University of Goias, University of Pisa, The Wistar Institute, CeMM, IRB Barcelona, EMBL-EBI) to perform high impact research. We also highly value the contributions of the more than 50 volunteers who have dedicated their time and expertise to Ersilia in the last four years, independently or via programs from Harvard University (Tech4SocialGood), UC Berkeley, the Good Data Institute, GitHub, Atlassian, Splunk and Digital Ocean.

Strategic Priorities

Priority 1: To build the reference resource for AI/ML research in infectious diseases

Goals:

Increase the number of AI models available through the Hub. Currently, Ersilia offers over 180 AI/ML models developed by third parties (75%) and by Ersilia (25%), related to bioactivity prediction, anticipation of toxicity, chemical properties, molecular descriptors, chemical space sampling and generative AI. Broadly speaking, Ersilia contains tools to annotate drug candidates with predicted properties and activities, explore the diversity and scope of chemical libraries, and sample the chemical space in search of new chemical matter. Our focus is to have a comprehensive set of activity prediction models and generative AI frameworks for chemistry, all related to anti-infectives drug discovery.
Improve accessibility and documentation of the Ersilia Model Hub. The Ersilia Model Hub is offered through a downloadable Command Line Interface (CLI) and as a Python Package (PyPi). Users can learn more about how to use Ersilia in our documentation (Ersilia Book). We aim to (a) consolidate the CLI in a release associated with a scientific publication, (b) clearly document all of Ersilia’s functions both for users and developers, and (c) continue the development of an LLM-based Ersilia assistant to facilitate user interaction with the platform.
Provide an online version of the Ersilia Model Hub. To bypass low computational capacity at partner organisations and facilitate its usage, we aim to deploy a selection of AI/ML models for online inference through the AWS cloud.
Complete an effort to pre-compute and store frequent user queries. We aim to save computational resources, provide faster service and enable large scale studies of the chemical space by persistently storing calculations in a fully open fashion.

KPIs:

Number and provenance of models available through the Hub.
Number of scientific publications citing the Hub and demonstrating its usage.
Number of users.
Number of contributors, understood as developers (Ersilia employees, interns, volunteers and open source contributors) as well as model donors.
Number of datapoints stored in our database.

Resources needed:

Personnel. The maintenance and extension of the Ersilia Model Hub requires two software engineers dedicated full-time to this task, and we estimate to engage 5-10 interns per year and an equal number of open source contributors. To expand our community engagement activities, Ersilia would require a project manager or community builder responsible for communicating with, maintaining and organising our open source community.
Infrastructure. Ersilia relies on free and open solutions for the majority of our infrastructural needs. The Ersilia Model Hub is hosted in a GitHub repository and released via PyPi and Conda. We benefit from a pro-bono plan to run GitHub Actions workflows that ensure continuous testing of the platform. In addition, Splunk has generously contributed a Splunk Enterprise license that allows us to monitor model usage and produce monthly reports. For cloud deployment solutions, we currently benefit from pro-bono credits from Digital Ocean, and we are building an AWS infrastructure. The cost of the AWS infrastructure will be covered by grants and pro-bono AWS credits.
Funding. Open source software has been traditionally difficult to maintain. To achieve our goals, Ersilia must raise funding to support the hiring of the detailed personnel as well as cover the costs of the backend infrastructure.
Data. AI model building is data intensive and most drug discovery-related data is locked in IP-protected databases. We liaise with organisations willing to share their data so we can build new models, as well as establish pipelines for automatically curating open data available in repositories and publications. Examples of such organisations would be the Swiss TPH, MMV and the Seattle Children’s Hospital.

Priority 2: To focus on novel therapeutic opportunities and targets for understudied diseases

Goals:

Identify new targets for orphan neglected tropical diseases using an AI-first approach.
Lead the adoption of targeted protein degradation in the field of antimicrobial drug discovery.
Explore the chemical space associated with antimicrobial drug discovery, with a special focus on natural product (-like) compounds.

KPIs:

Number of diseases investigated in our research programs.
Number of validated targets identified.
Number of collaborations with research groups with experimental capacity to test our hypotheses.
Number of scientific publications and associated citations.

Resources needed:

Personnel. Dedicated, highly trained research personnel (PhD students and Postdoctoral researchers) will be required to perform such studies. We will hire at least one new postdoctoral researcher and offer a 4-year PhD position at Ersilia.
Infrastructure. GPU-powered resources (local and cloud) are required to perform these analyses. We will use Ersilia’s own internal resources as well as partner with universities and research centers with access to computer clusters that can fill in our needs.
Funding. Dedicated research funding from governmental and international sources (Spanish Ministry of Science, Innovation and Universities, Horizon EU, Wellcome Trust, NIH and others) will be applied to cover the costs of research and the funding for specific research positions (PhD student stipends, Postdoctoral salaries).

Priority 3: To provide long-term training opportunities for scientists in the Global South

Goals:

Host at least one in-person workshop each year in a LMIC. Our in-person workshops have a duration of one week and offer an introduction to basic AI/ML concepts as well as hands-on practice on the application of AI models to drug discovery.
Run the first cohort of the AI incubator program. We will host at least 6 research groups working on infectious disease drug discovery and accompany them in their journey to incorporate data science and AI into their research programs. While the first cohort will be focused on drug discovery, we aim to expand to other research areas such as biomarker discovery for future cohorts.
Provide mentorship programs for underrepresented groups in STEM. Introducing the next generation of software developers and data scientists to the realm of open source research software and biomedical sciences allows the growth of local talent that can be incorporated into the working force for science and technology in LMIC. We achieve that via paid internships and mentoring of MSc theses for computer scientists.

KPIs:

Number of scientists trained in AI/ML for biomedical research.
Number of countries targeted by our training programmes.
Number of interns trained.
Number of MSc theses co-directed by Ersilia.

Resources needed:

Personnel. To achieve our training and capacity building objectives we need a combination of skills. On one hand, Ersilia’s researchers must be ready to deliver training, provide online mentorship and engage with the trainees projects, while Ersilia’s software engineers must ensure our technology responds to the users needs and is accessible to them. Finally, administrative personnel (program manager or similar) is key to ensure the success of larger programs like the AI Incubator.
Funding. To achieve our training goals we will rely mostly on philanthropic support from organisations that engage in science for development and open source. To continue our partnership with Outreachy, we need to raise 10.000 USD per each Outreachy intern (8000 USD per salary and 2000 USD to contribute to Outreachy’s overhead costs).

Priority 4: To continue building a community that maximises our impact and ensures our sustainability

Goals:

Establish long-term partnerships with local Global South research institutions to further support outstanding scientists in their home countries who can benefit from Ersilia’s expertise and become stakeholders in the development of data science facilities in their institutions.
Grow a solid network of collaborators and supporters locally in Barcelona. For the first four years, Ersilia has been essentially a remote organization. It is now time to participate in the research ecosystem in our home city, and in Europe more broadly, by engaging in debate, conferences, and activities happening locally.
Build a strong diversity and inclusion work environment. At any given time, between 5 to 20 volunteers and contributors might be working with Ersilia in addition to our full-time staff. These volunteers come from various different countries, backgrounds and expertises, and they are a fundamental part of our work. Ensuring they all feel welcome and safe at Ersilia is paramount to our success.
Expand Ersilia’s advisory board, currently composed of Ersilia’s UK former Trustees, to support our decision-making processes and ensure Ersilia is an organization that answers the real needs of those whom it serves.
Diversify Ersilia’s funding sources. To ensure long term sustainability, we will continue combining our fundraising efforts with grant applications and research funding. We will also expand our donation programs in Spain and the US.
Build internal capacity with key hires in technical and operational roles. To fulfill all the above priorities, increasing Ersilia’s full-time employees with the appropriate skill sets will be required.

KPIs:

Number of partnerships with research institutions, in Africa and elsewhere.
Number of full-time employees at Ersilia.
Number of Advisors and diversity of expertises available.
Funding raised to advance Ersilia’s mission.

Resources needed:

Personnel. The strengthening of Ersilia’s Governance, Community Building and Fundraising will be led by Ersilia’s co-founders Miquel and Gemma with the support of the Board of Trustees.

Financial and resource plan

Open Source funding. Open source funding for research software from tech organisations (Mozilla Foundation, Splunk, GitHub, Google and others) and governmental initiatives (FAIR-Impact, Horizon EU, OSCARS) will be devoted to achieving Priority 1. The goal is to raise 0.5M USD in the next three years in that area.
Research funding. Scientific programs from the European Union, the United Kingdom, Spain and the United States will be the main target to support the work of Priority 2, along with science philanthropy funders like the Chan Zuckerberg Initiative or the Bill and Melinda Gates Foundation. We aim to secure a large, multi-year grant to expand our scientific program as well as secure funding for academic positions within Ersilia.
Philanthropy. Traditional philanthropy and accelerator-based funding will be devoted to achieving Priorities 3 and 4. The goal is to secure a main funder for the Ersilia AI Incubator by the end of 2025, and to participate in an accelerator program for scale-up stage initiatives by 2026.

Appendix

An up-to-date version of the statistics referred to in this report are available under a CC-BY-4.0 License in the Ersilia Stats repository.

PreviousErsilia privacy notice NextErsilia, the Invisible City

Last updated 1 day ago

Was this helpful?