Biology subjects

Castro, K.

Publications and source records attributed to Castro, K..

4 recordsLinked to original sources

bioRxiv2026

Contrastive learning for antibody-antigen sequence-to-specificity prediction

Predicting which antibodies bind to which antigens directly from primary amino acid sequences remains a major challenge, as no current method can reliably determine this specificity at both a repertoire and proteome scale. Structure-based protein design frameworks can propose antibody binders to specified antigenic epitopes, but they do not solve the "sequence-to-specificity" task of mapping antibodies to cognate epitopes, and vice versa. Here, we introduce CALM (Cross-attention Adaptive Immune Receptor-Antigen Language Model), a dual-encoder plus cross-attentive decoder architecture that treats antibody-antigen recognition as molecular translation. Using contrastive learning, antigen and antibody encoders learn a shared embedding space that aligns cognate epitope-paratope binding pairs. CALM-1.0 is trained and evaluated on 4,138 curated antibody-antigen pairs obtained from the PDB-derived structural antibody database (SAbDab). On a leakage-controlled test split drawn from sequence clusters at 80% identity and unseen during training, CALM-1.0 achieves a mean top-1 retrieval (R@1) of 7%, with consistent performance across both directions (Ab[->]Ag and Ag[->]Ab). CALM establishes a foundation for bidirectional antibody-antigen sequence-to-specificity prediction with the potential to unify retrieval and generative design.

immunology↗

bioRxiv2024

Autoaggregation in Streptococcus intermedius is driven by the Pel polysaccharide

The Streptococcus Milleri Group (SMG) comprising of Streptococcus intermedius, anginosus and constellatus are commensal bacteria commonly found in healthy individuals. These bacteria are increasingly being recognized as opportunistic pathogens that can cause purulent infections at sterile body sites and have also been identified in the sputum of individuals with cystic fibrosis. Although the mechanisms of conversion to opportunistic pathogens are not well understood, auto-aggregation is a key driver of biofilm adhesion and cohesion in many Streptococci and Staphylococci. Here, we identify a gene cluster in the S. intermedius genome with significant homology to the pel operons in Bacillus cereus and Pseudomonas aeruginosa, which are required for Pel exopolysaccharide production and biofilm formation in these species. Characterization of a panel of clinical S. intermedius strains identified a range of aggregating phenotypes. Analysis of the pel operon in the hyper-aggregating C1365 strain revealed that each of the canonical pelDEADAFG genes, but not the four additional genes are required for aggregation. Further, we demonstrate that C1365 produces a GalNAc-rich exopolysaccharide and that aggregates can be disrupted by the 1,4 N-acetylgalactosaminidases, PelA and Sph3, but not other glycoside hydrolases, proteinase K or DNase I. Using an abscess model of mouse infection, we show that Pel driven aggregation leads to longer lasting infections, and that lack of Pel allows for the bacteria to be cleared more effectively. The polymer also affects how the bacteria interacts with the host immune system. Collectively, our data suggest that the pel operon has relevancy to S. intermedius pathogenicity.

microbiology↗

bioRxiv2020

A Microfluidic Platform for Functional Testing of Cancer Drugs on Intact Tumor Slices

Present approaches to assess cancer treatments are often inaccurate, costly, and/or cumbersome. Functional testing platforms that use live tumor cells are a promising tool both for drug development and for identifying the optimal therapy for a given patient, i.e. precision oncology. However, current methods that utilize patient-derived cells from dissociated tissue typically lack the microenvironment of the tumor tissue and/or cannot inform on a timescale rapid enough to guide decisions for patient-specific therapy. We have developed a microfluidic platform that allows for multiplexed drug testing of intact tumor slices cultured on a porous membrane. The device is digitally-manufactured in a biocompatible thermoplastic by laser-cutting and solvent bonding. Here we describe the fabrication process in detail, we characterize the fluidic performance of the device, and demonstrate on-device drug-response testing with tumor slices from xenografts and from a patient colorectal tumor.

bioengineering↗

bioRxiv2020

Multiplexed drug testing of tumor slices using a microfluidic platform

Current methods to assess the drug response of individual human cancers are often inaccurate, costly, or slow. Functional approaches that rapidly and directly assess the response of patient cancer tissue to drugs or small molecules offer a promising way to improve drug testing, and have the potential to identify the best therapy for individual patients. We developed a digitally-manufactured microfluidic platform for multiplexed drug testing of intact cancer slice cultures, and demonstrate the use of this platform to evaluate drug responses in slice cultures from human glioma xenografts and patient tumor biopsies. This approach retains much of the tissue microenvironment and can provide results rapidly enough, within days of surgery, to guide the choice of effective initial therapies. Our results establish a useful preclinical platform for cancer drug testing and development with the potential to improve cancer personalized medicine.

bioengineering↗