ImmPort, toward repurposing of open access immunological assay data for translational and clinical research

Sanchita Bhattacharya, Patrick Dunn, Cristel G. Thomas, Barry Smith, Henry Schaefer et al. (+8)•View original

HighlightsAccessiblecoral voice

If you’ve ever tried to make sense of the immune system, you know the data come fast and messy. Every lab measures different things in different ways. It’s like trying to compare recipes when half are in cups, half in grams, and some just say "a pinch." ImmPort is the big shared pantry that fixes that. It’s an open repository where immunology data live in one place, described the same way, and ready to be reused. By early 2018, Bhattacharya and colleagues reported three hundred and nine studies covering fifty thousand one hundred and eighty human and animal subjects inside ImmPort. That’s not a pile of papers. That’s subject-level data you can actually analyze. Here’s how it works in practice. ImmPort is an ecosystem with four parts: Private Data, Shared Data, Data Analysis, and Resources. New datasets land in Private Data, where providers curate and control access. Once they’re cleaned up and de-identified, they’re published to Shared Data for anyone to explore. If you want to analyze without writing code, Data Analysis offers ImmPort Galaxy, a point-and-click portal built on the Galaxy framework to run open-source cytometry tools. And when you need to learn a method or find a reference dataset, Resources is the help desk—tutorials, examples, and tools in one spot. Finding the right data is half the battle, so Shared Data includes a searchable catalog. You can look by study, biomarker, experiment type, or lab test, then download subject-level files as simple tab-separated tables or MySQL databases. There are application programming interfaces, or APIs, if you need custom extracts. Upload templates nudge contributors toward consistent naming. Privacy isn’t an afterthought: data are de-identified to meet Health Insurance Portability and Accountability Act rules, and especially sensitive material can be routed to controlled-access archives like the Database of Genotypes and Phenotypes, or dbGaP, or the Sequence Read Archive. Standard language is what makes all of this click. ImmPort annotates datasets with controlled vocabularies—think shared dictionaries—so the same cell type or disease means the same thing across studies. That includes the Cell Ontology, Disease Ontology, the Ontology for Biomedical Investigations, Protein Ontology, and Vaccine Ontology, with the Medical Dictionary for Regulatory Activities, or MedDRA, for adverse events and the National Cancer Institute Thesaurus for clinical trial terms. They even built the Antibody Ontology, called AntiO, from hands-on curation so you can ask very specific questions like, "Show me all PE-labeled anti-CD25 antibodies, with vendor and catalog number, and the studies that used them." For signaling molecules like cytokines and chemokines, a dedicated registry ties together names and synonyms from sources like UniProt and the National Center for Biotechnology Information Gene database so you don’t miss data just because someone used an alternate label. And if you want to go beyond search, there are tools to lower the barrier to cross-study analysis. MetaCyto lets you do automated meta-analysis of cytometry—those experiments that count and characterize cells—even when panels and batches differ across studies. RImmPort brings a clinical data standard called the Clinical Data Interchange Standards Consortium, or CDISC, into R, so you can merge and analyze studies with fewer headaches. Under the hood, ImmPort has grown deep: one thousand three hundred and sixty-nine experiments, two hundred and thirty-six lab test panels, and four hundred and forty-nine assessments were available by 2018, and data from sixty-five clinical trials had been shared. That breadth matters when you’re looking for patterns that only appear at scale. What does reuse look like in the wild? Khatri and colleagues pulled gene-expression signatures to predict who would respond to influenza vaccines—an example of turning scattered measurements into a practical forecast. Nasrallah’s reanalysis of the Rituximab RAVE trial—Rituximab in Anti-Neutrophil Cytoplasmic Antibody Associated Vasculitis—in anti-neutrophil cytoplasmic antibody associated vasculitis surfaced early granulocyte subsets that weren’t the headline in the original paper. Strauli and Hernandez traced how antibody repertoires evolve over time. Projects like the ten thousand Immunomes and immuneXpresso show how, once data are standardized and shared, entirely new maps of human immunity emerge. And ImmPort sits at the hub, also powering consortia such as the Accelerating Medicines Partnership in rheumatoid arthritis and lupus, the Human Immunology Project Consortium, ImmuneSpace, the National Cancer Institute Oncology Model Forum, and the March of Dimes Prematurity Centers. If you remember three things, make it these. First, ImmPort turns scattered immunology data into a searchable, reusable commons, aligned with FAIR principles—findable, accessible, interoperable, reusable—so you spend less time wrangling and more time discovering. Second, it pairs that commons with practical tools—ImmPort Galaxy, MetaCyto, RImmPort, and AntiO—that make cross-study analysis doable. Third, this isn’t theoretical; hundreds of studies and tens of thousands of subjects are already in play, and they’ve already powered new insights. In a field where timing matters, having a shared, well-labeled pantry can change what we’re able to cook up.

It’s an open repository where immunology data live in one place, described the same way, and ready to be reused. By early 2018, Bhattacharya and colleagues reported three hundred and nine studies covering fifty thousand one hundred and eighty human and animal subjects inside ImmPort. That’s not a pile of papers. That’s subject-level data you can actually analyze.

Here’s how it works in practice. ImmPort is an ecosystem with four parts: Private Data, Shared Data, Data Analysis, and Resources. New datasets land in Private Data, where providers curate and control access.

Once they’re cleaned up and de-identified, they’re published to Shared Data for anyone to explore. If you want to analyze without writing code, Data Analysis offers ImmPort Galaxy, a point-and-click portal built on the Galaxy framework to run open-source cytometry tools. And when you need to learn a method or find a reference dataset, Resources is the help desk—tutorials, examples, and tools in one spot.

Finding the right data is half the battle, so Shared Data includes a searchable catalog. You can look by study, biomarker, experiment type, or lab test, then download subject-level files as simple tab-separated tables or MySQL databases. There are application programming interfaces, or APIs, if you need custom extracts.

Upload templates nudge contributors toward consistent naming. Privacy isn’t an afterthought: data are de-identified to meet Health Insurance Portability and Accountability Act rules, and especially sensitive material can be routed to controlled-access archives like the Database of Genotypes and Phenotypes, or dbGaP, or the Sequence Read Archive.

Standard language is what makes all of this click. ImmPort annotates datasets with controlled vocabularies—think shared dictionaries—so the same cell type or disease means the same thing across studies. That includes the Cell Ontology, Disease Ontology, the Ontology for Biomedical Investigations, Protein Ontology, and Vaccine Ontology, with the Medical Dictionary for Regulatory Activities, or MedDRA, for adverse events and the National Cancer Institute Thesaurus for clinical trial terms.

They even built the Antibody Ontology, called AntiO, from hands-on curation so you can ask very specific questions like, "Show me all PE-labeled anti-CD25 antibodies, with vendor and catalog number, and the studies that used them." For signaling molecules like cytokines and chemokines, a dedicated registry ties together names and synonyms from sources like UniProt and the National Center for Biotechnology Information Gene database so you don’t miss data just because someone used an alternate label.

And if you want to go beyond search, there are tools to lower the barrier to cross-study analysis. MetaCyto lets you do automated meta-analysis of cytometry—those experiments that count and characterize cells—even when panels and batches differ across studies. RImmPort brings a clinical data standard called the Clinical Data Interchange Standards Consortium, or CDISC, into R, so you can merge and analyze studies with fewer headaches.

Under the hood, ImmPort has grown deep: one thousand three hundred and sixty-nine experiments, two hundred and thirty-six lab test panels, and four hundred and forty-nine assessments were available by 2018, and data from sixty-five clinical trials had been shared. That breadth matters when you’re looking for patterns that only appear at scale.

What does reuse look like in the wild? Khatri and colleagues pulled gene-expression signatures to predict who would respond to influenza vaccines—an example of turning scattered measurements into a practical forecast. Nasrallah’s reanalysis of the Rituximab RAVE trial—Rituximab in Anti-Neutrophil Cytoplasmic Antibody Associated Vasculitis—in anti-neutrophil cytoplasmic antibody associated vasculitis surfaced early granulocyte subsets that weren’t the headline in the original paper.

Strauli and Hernandez traced how antibody repertoires evolve over time. Projects like the ten thousand Immunomes and immuneXpresso show how, once data are standardized and shared, entirely new maps of human immunity emerge. And ImmPort sits at the hub, also powering consortia such as the Accelerating Medicines Partnership in rheumatoid arthritis and lupus, the Human Immunology Project Consortium, ImmuneSpace, the National Cancer Institute Oncology Model Forum, and the March of Dimes Prematurity Centers.

If you remember three things, make it these. First, ImmPort turns scattered immunology data into a searchable, reusable commons, aligned with FAIR principles—findable, accessible, interoperable, reusable—so you spend less time wrangling and more time discovering. Second, it pairs that commons with practical tools—ImmPort Galaxy, MetaCyto, RImmPort, and AntiO—that make cross-study analysis doable.

Third, this isn’t theoretical; hundreds of studies and tens of thousands of subjects are already in play, and they’ve already powered new insights. In a field where timing matters, having a shared, well-labeled pantry can change what we’re able to cook up.