Linke Li

Can infertility offer clues to later-life 🫀 biology? Work led by ⭐️ @mghmedres Micayla Flores now in @JACCJournals #Advances: Infertility was associated with circulating proteins related to inflammation and tissue remodeling in @WHIStudy and @uk_biobank.

Excited to share our new publication: "Exome-wide association study for blood lipids in 1,158,017 individuals from diverse populations." nature.com/articles/s4158… This study explores how multi-ancestry DNA sequencing has identified rare coding variants that influence blood lipids and atherosclerotic risk. Highlights include: • Discovery of rare coding variants across more than 1 million participants • Identification of novel lipid-associated genes • Insights into variant annotation and recessive associations • Potential therapeutic targets for coronary artery disease • The power of global biobank collaboration I extend my gratitude to all collaborators, participants, and consortium teams who made this research possible. I am proud to be part of this effort. Special thanks to my mentors, Drs. Yan Sun, Peter Wilson, @patrick_ellinor, and @pnatarajanmd, for their guidance and support throughout this work! For more details, see the previous thread: x.com/skoyamamd/stat…

🎉Happy to share our (@LinkeLi_MGH @tigerstatdoc @M_Mesbah_Uddin @abhinrl @pnatarajanmd me) collaborative article with Reuben & Santhosh's team at @IUMedSchool, just out in @BloodPortfolio! Clonal hematopoiesis of indeterminate potential (CHIP)🧬is linked to multiple organ systems — heart, liver, brain — what about the colon? And if there is a causal link, can we do something about it? Using Dnmt3a-mutant mouse 🐭models of colitis and human🧓data from UK Biobank & Mass General Brigham, we show that CHIP increases inflammatory bowel disease (IBD) risk; further, APE1/Ref-1 inhibitor APX3330 can ameliorate CHIP-driven IBD, potentially breaking the CHIP-disease link. 🔗sciencedirect.com/science/articl…

🎉Check out our preprint led by @LinkeLi_MGH @_ahmedmalaa with @pnatarajanmd : ) One of the biggest hurdles in proteomics is the non-replication of associations due to limited cross-platform correlation. We tackle this with ML models that bridge the two major ones, SomaScan and Olink, plus a tiered protein reliability system to guide which signals to trust. We validate against gold-standard measurements, AlphaFold 3 predicted structures, and 3 real-world applications, including replication of published dementia🧠and heart failure🫀associations, showing markedly improved cross-platform replicability. biorxiv.org/content/10.648…

Plus a per-protein fidelity index, validated vs. ELISA & interpreted w/ AlphaFold3. We show fidelity-informed + imputed data boost replication, recover SomaScan signals invisible in UKB's Olink, and improve overlap-protein replication.

🚨Excited to share our new preprint!🚨 A big headache in proteomics: findings don't replicate across SomaScan & Olink. ML models trained on 5K paired MESA samples can narrow the gap! biorxiv.org/content/10.648…

Our collaboration with R. Kumar & R. Kapur @IUMedSchool shows how clonal hematopoiesis driver mutations influence the likelihood of developing inflammatory bowel disease ashpublications.org/blood/article-… @BloodPortfolio L. Li @ZhiYu_ACGT @M_Mesbah_Uddin @tigerstatdoc

Join our CTEU Seminar on 2/18 at 9 AM. @LinkeLi_MGH from MGH will give a talk entitled “Machine learning cross-platform proteomic imputation enables protein quality scoring and replication of epidemiological associations.” Chat with us at mghcteu.org to join.