Tu186 - Data Integration Across Different Mass Cytometers: Implication in Data Sharing and Re-use for Discovery

Martin Wasser – Translational Immunology Institute, SingHealth Duke-NUS Academic Medical Centre; Katherine Nay Yaung – Translational Immunology Institute, SingHealth Duke-NUS Academic Medical Centre; Dianyan Guo – Translational Immunology Institute, SingHealth Duke-NUS Academic Medical Centre; Su Li Poh – Translational Immunology Institute, SingHealth Duke-NUS Academic Medical Centre; Judith Wong – Translational Immunology Institute, SingHealth Duke-NUS Academic Medical Centre; Thaschawee Arkachaisri – KK Women’s and Children’s Hospital, Singapore; Joo Guan Yeo – Translational Immunology Institute, SingHealth Duke-NUS Academic Medical Centre; Salvatore Albani – Translational Immunology Institute, SingHealth Duke-NUS Academic Medical Centre

Abstract Text: Recently, emphasis has been placed on the acquisition of Big Data using multi-parametric cytometer to study complex diseases, and their sharing through web portals such as ImmPort for integration and reuse for novel discovery. However, issues such as batch effects hinder this process and strategies for its mitigation lacking.
To address this unmet need, we interrogated multiple vials of cryopreserved peripheral blood mononuclear cells (PBMCs) from a single preparation of an adult healthy blood donor (technical replicate, n=10) and healthy volunteers (children and adults, n=52) over six months using two different mass cytometers (Helios and XT). A standardised operating procedure (SOP) for thawing, stimulation and staining was performed. Batch-scaling, clustering and depiction of the degree of similarities of the cell frequencies between the technical replicates using t-SNE and PCA were done.
With batch-scaling, our technical replicates (n=10) clustered in close proximity with no data segregation by cytometer (Helios or XT) used. Additionally, the frequencies of major immune lineages such as CD4+ (Coefficient of variation, CoV: 8.0% ) and CD8+ (CoV: 4.6%) across both cytometers showed an acceptable CoV, indicative of data reproducibility. Importantly, our healthy donors’ data (n=52) segregated by age with a distinct age gradient with the healthy adults clustered closer to the technical replicates.
Our findings suggest that the concerns regarding data integration from various cytometers can be mitigated by using batch-scaling with a SOP to handle experimental batch effects. Translationally, repurposing and integrating Big Data to boost statistical power with larger sample sizes can hasten novel discovery.