Albert Henry
@alberthenry.bsky.social
MD 🇮🇩 | MSc 🇬🇧 | PhD 🇬🇧
Research officer - Garvan Institute of Medical Research 🇦🇺
Honorary research fellow - University College London 🇬🇧
Interested in genetics, phenotypes, and anything in-between
Research officer - Garvan Institute of Medical Research 🇦🇺
Honorary research fellow - University College London 🇬🇧
Interested in genetics, phenotypes, and anything in-between
Big thanks to co-first author Anne Senabouth, supervisor Joseph Powell, co-authors Rika Tyebally @blakebowen.bsky.social @lawrencehuang.bsky.social Jayden Fan @petercallen.bsky.social @anglixue.bsky.social @htanudisastro.bsky.social @annasecuomo.bsky.social @dgmacarthur.bsky.social, and many others!
September 1, 2025 at 4:31 AM
Big thanks to co-first author Anne Senabouth, supervisor Joseph Powell, co-authors Rika Tyebally @blakebowen.bsky.social @lawrencehuang.bsky.social Jayden Fan @petercallen.bsky.social @anglixue.bsky.social @htanudisastro.bsky.social @annasecuomo.bsky.social @dgmacarthur.bsky.social, and many others!
10. Last but not least, this study would not be possible without massive contributions from all the co-authors and the wider TenK10K team. If you like to learn more about TenK10K, check out our other studies:
tinyurl.com/tenk10k-flagship
tinyurl.com/tenk10k-repeats
tinyurl.com/tenk10k-multiome
tinyurl.com/tenk10k-flagship
tinyurl.com/tenk10k-repeats
tinyurl.com/tenk10k-multiome
September 1, 2025 at 4:31 AM
10. Last but not least, this study would not be possible without massive contributions from all the co-authors and the wider TenK10K team. If you like to learn more about TenK10K, check out our other studies:
tinyurl.com/tenk10k-flagship
tinyurl.com/tenk10k-repeats
tinyurl.com/tenk10k-multiome
tinyurl.com/tenk10k-flagship
tinyurl.com/tenk10k-repeats
tinyurl.com/tenk10k-multiome
9. We hope our study offers a cell type-resolved map for causal inference of gene expression on complex traits to help understand disease mechanisms and guide drug development.
There are still plenty to unpack. We encourage reading the preprint and would love to hear feedback!
There are still plenty to unpack. We encourage reading the preprint and would love to hear feedback!
September 1, 2025 at 4:31 AM
9. We hope our study offers a cell type-resolved map for causal inference of gene expression on complex traits to help understand disease mechanisms and guide drug development.
There are still plenty to unpack. We encourage reading the preprint and would love to hear feedback!
There are still plenty to unpack. We encourage reading the preprint and would love to hear feedback!
8. We found 116 genes associated with Crohn’s disease show differential expression in equivalent colon tissue cell types sampled from healthy and diseased individuals in an external dataset. This includes ZBTB38, a candidate susceptibility gene implicated in recent GWAS.
September 1, 2025 at 4:31 AM
8. We found 116 genes associated with Crohn’s disease show differential expression in equivalent colon tissue cell types sampled from healthy and diseased individuals in an external dataset. This includes ZBTB38, a candidate susceptibility gene implicated in recent GWAS.
7. We highlight an interesting example of cell type-specific causal effect of NCF4 gene in Crohn’s disease, which shows a protective (-) effect in B naive and risk-increasing (+) effect in B memory, implicating context-specific regulation that can be resolved using sc-eQTL MR.
September 1, 2025 at 4:31 AM
7. We highlight an interesting example of cell type-specific causal effect of NCF4 gene in Crohn’s disease, which shows a protective (-) effect in B naive and risk-increasing (+) effect in B memory, implicating context-specific regulation that can be resolved using sc-eQTL MR.
6. Drugs targeting gene-trait associations identified through sc-eQTL MR in this study are 3.3 times more likely to have secured regulatory approval. Among these were major targets such as JAK2 & TNF for Crohn’s disease, APP for Alzheimer’s disease, and GLP1R for type 2 diabetes.
September 1, 2025 at 4:31 AM
6. Drugs targeting gene-trait associations identified through sc-eQTL MR in this study are 3.3 times more likely to have secured regulatory approval. Among these were major targets such as JAK2 & TNF for Crohn’s disease, APP for Alzheimer’s disease, and GLP1R for type 2 diabetes.
5. We found different polygenic enrichment patterns amongst dendritic cell (DC) subtypes: Crohn’s disease enrichment were found in cDC1, cDC2, and ASDC subtypes, and COVID-19 found in pDC only - consistent with its function for rapid interferon signalling in viral infection.
September 1, 2025 at 4:31 AM
5. We found different polygenic enrichment patterns amongst dendritic cell (DC) subtypes: Crohn’s disease enrichment were found in cDC1, cDC2, and ASDC subtypes, and COVID-19 found in pDC only - consistent with its function for rapid interferon signalling in viral infection.
4. Through single-cell Disease Relevance Score (scDRS) analysis, we found that peripheral immune cells are enriched for polygenic signature of most complex traits, implicating widespread pleiotropy beyond immune function and peripheral blood composition.
September 1, 2025 at 4:31 AM
4. Through single-cell Disease Relevance Score (scDRS) analysis, we found that peripheral immune cells are enriched for polygenic signature of most complex traits, implicating widespread pleiotropy beyond immune function and peripheral blood composition.
3. We identified 190,449 gene-trait associations, including 34% not found by gene-level analysis of GWAS data, and 61% not found by MR using whole blood eQTL. Associations found only by sc-eQTL MR are often restricted to fewer cell types, implicating cell type specificity.
September 1, 2025 at 4:31 AM
3. We identified 190,449 gene-trait associations, including 34% not found by gene-level analysis of GWAS data, and 61% not found by MR using whole blood eQTL. Associations found only by sc-eQTL MR are often restricted to fewer cell types, implicating cell type specificity.
2. Our study presents a catalogue of cell type-specific causal effects of gene expression on 53 diseases (8,672 genes), and 31 biomarker traits (16,085 genes) across 28 peripheral immune cell types identified using Mendelian randomisation (MR) with sc-eQTL genetic instruments.
September 1, 2025 at 4:31 AM
2. Our study presents a catalogue of cell type-specific causal effects of gene expression on 53 diseases (8,672 genes), and 31 biomarker traits (16,085 genes) across 28 peripheral immune cell types identified using Mendelian randomisation (MR) with sc-eQTL genetic instruments.
Lastly, it goes without saying that it takes a village to publish this study. I'd like to take this opportunity to thank my previous PhD and postdoc advisor at UCL, Dr. Tom Lumbers who led this project, friends and collaborators within the HERMES Consortium, and all the study participants.
END/
END/
March 4, 2025 at 11:29 AM
Lastly, it goes without saying that it takes a village to publish this study. I'd like to take this opportunity to thank my previous PhD and postdoc advisor at UCL, Dr. Tom Lumbers who led this project, friends and collaborators within the HERMES Consortium, and all the study participants.
END/
END/
BONUS for those who scroll long enough to find this:
We also have an online supplementary information with more details on:
1. GWAS QC pipeline
2. Locus zoom, gene prioritisation, cross-trait association, and study-level association for each heart failure locus
hermes2-supp-note.netlify.app
14/
We also have an online supplementary information with more details on:
1. GWAS QC pipeline
2. Locus zoom, gene prioritisation, cross-trait association, and study-level association for each heart failure locus
hermes2-supp-note.netlify.app
14/
Genome-wide association study meta-analysis provides insights into the etiology of heart failure and its subtypes
hermes2-supp-note.netlify.app
March 4, 2025 at 11:29 AM
BONUS for those who scroll long enough to find this:
We also have an online supplementary information with more details on:
1. GWAS QC pipeline
2. Locus zoom, gene prioritisation, cross-trait association, and study-level association for each heart failure locus
hermes2-supp-note.netlify.app
14/
We also have an online supplementary information with more details on:
1. GWAS QC pipeline
2. Locus zoom, gene prioritisation, cross-trait association, and study-level association for each heart failure locus
hermes2-supp-note.netlify.app
14/
We have also released the GWAS summary statistics for browsing and download via the CVD Knowledge Portal:
* Mixed-ancestry meta-analysis:
cvd.hugeamp.org/dinspector.h...
* European ancestry meta-analysis:
cvd.hugeamp.org/dinspector.h...
13/
* Mixed-ancestry meta-analysis:
cvd.hugeamp.org/dinspector.h...
* European ancestry meta-analysis:
cvd.hugeamp.org/dinspector.h...
13/
Datasets | Common Metabolic Diseases Knowledge Portal
cvd.hugeamp.org
March 4, 2025 at 11:29 AM
We have also released the GWAS summary statistics for browsing and download via the CVD Knowledge Portal:
* Mixed-ancestry meta-analysis:
cvd.hugeamp.org/dinspector.h...
* European ancestry meta-analysis:
cvd.hugeamp.org/dinspector.h...
13/
* Mixed-ancestry meta-analysis:
cvd.hugeamp.org/dinspector.h...
* European ancestry meta-analysis:
cvd.hugeamp.org/dinspector.h...
13/
We described some more analyses in the paper that are not covered here; including genetic architecture, heritability, polygenic risk score, finemapping and pathway enrichment.
Do have a read if you find our paper interesting, and let us know if you have any feedback!
12/
Do have a read if you find our paper interesting, and let us know if you have any feedback!
12/
March 4, 2025 at 11:29 AM
We described some more analyses in the paper that are not covered here; including genetic architecture, heritability, polygenic risk score, finemapping and pathway enrichment.
Do have a read if you find our paper interesting, and let us know if you have any feedback!
12/
Do have a read if you find our paper interesting, and let us know if you have any feedback!
12/
We performed genetic correlation (rg) and Mendelian randomisation analyses to distinguish between shared genetics and causal relationships. This is most apparent in CAD and ni-HF, which shows positive rg without causal effect. Interestingly, T2D shows this pattern on all HF subtypes.
11/
11/
March 4, 2025 at 11:29 AM
We performed genetic correlation (rg) and Mendelian randomisation analyses to distinguish between shared genetics and causal relationships. This is most apparent in CAD and ni-HF, which shows positive rg without causal effect. Interestingly, T2D shows this pattern on all HF subtypes.
11/
11/
We further explored the extent of pleiotropic effects in HF loci on risk factors and diseases associated with HF. Through colocalisation analysis, we found that HF shared causal genetic variants with at least one of 22 other traits at 42 loci.
10/
10/
March 4, 2025 at 11:29 AM
We further explored the extent of pleiotropic effects in HF loci on risk factors and diseases associated with HF. Through colocalisation analysis, we found that HF shared causal genetic variants with at least one of 22 other traits at 42 loci.
10/
10/
The identified genotype-phenotype clusters provide insights into etiological modules underlying HF pathology, e.g. cluster 1: ischaemic & major cardiovascular disorders, cluster 2: arrythmia & cardiomyopathies, cluster 4: hypertension, cluster 5: metabolic disorders.
9/
9/
March 4, 2025 at 11:29 AM
The identified genotype-phenotype clusters provide insights into etiological modules underlying HF pathology, e.g. cluster 1: ischaemic & major cardiovascular disorders, cluster 2: arrythmia & cardiomyopathies, cluster 4: hypertension, cluster 5: metabolic disorders.
9/
9/
Next, we characterised the downstream effect of lead variants in HF susceptibility loci on 294 human diseases in UK Biobank. We then used network analysis and community detection technique to identify 18 distinct genotype-phenotype clusters from these phenome-wide association results.
8/
8/
March 4, 2025 at 11:29 AM
Next, we characterised the downstream effect of lead variants in HF susceptibility loci on 294 human diseases in UK Biobank. We then used network analysis and community detection technique to identify 18 distinct genotype-phenotype clusters from these phenome-wide association results.
8/
8/
Using sn-RNAseq from 16 healthy & 28 failing heart donors, we found enrichment of cardiomyocyte genes. We also identified 53 GWAS genes that were differentially expressed in cardiac cell types, notably cardiomyocytes and fibroblasts.
7/
7/
March 4, 2025 at 11:29 AM
Using sn-RNAseq from 16 healthy & 28 failing heart donors, we found enrichment of cardiomyocyte genes. We also identified 53 GWAS genes that were differentially expressed in cardiac cell types, notably cardiomyocytes and fibroblasts.
7/
7/
Through heritability enrichment analysis, we found differential involvement of tissues across HF subtypes. Notably, whilst other HF subtypes were mostly enriched for genes that are more specifically expressed in cardiac tissues, ni-HFpEF was distinctly enriched for kidney and pancreas.
6/
6/
March 4, 2025 at 11:29 AM
Through heritability enrichment analysis, we found differential involvement of tissues across HF subtypes. Notably, whilst other HF subtypes were mostly enriched for genes that are more specifically expressed in cardiac tissues, ni-HFpEF was distinctly enriched for kidney and pancreas.
6/
6/
Integrating multiple gene prioritisation strategies, we shortlisted 142 candidate effector genes across 66 genetic loci for HF, and nominated the most likely effector gene for each locus. This includes IGFBP7 for HFpEF, which is linked to cardiomyocyte senescence and cardiac remodelling.
5/
5/
March 4, 2025 at 11:29 AM
Integrating multiple gene prioritisation strategies, we shortlisted 142 candidate effector genes across 66 genetic loci for HF, and nominated the most likely effector gene for each locus. This includes IGFBP7 for HFpEF, which is linked to cardiomyocyte senescence and cardiac remodelling.
5/
5/
We found 66 independent genetic loci associated with at least 1 HF phenotype, including 37 not previously linked to HF. Of note, 10 / 66 loci were identified in GWAS of ni-HF subtypes despite smaller N compared to HF-all; showing the importance of phenotype definition in a case-control GWAS
4/
4/
March 4, 2025 at 11:29 AM
We found 66 independent genetic loci associated with at least 1 HF phenotype, including 37 not previously linked to HF. Of note, 10 / 66 loci were identified in GWAS of ni-HF subtypes despite smaller N compared to HF-all; showing the importance of phenotype definition in a case-control GWAS
4/
4/