Biomedical subjects

Sai Zhang

Publications and source records attributed to Sai Zhang.

3 recordsLinked to original sources

PubMed2026

ProMeta: a meta-learning framework for robust disease diagnosis and prediction from plasma proteomics.

MOTIVATION: The plasma proteome offers a dynamic window of human health, capturing the real-time intersections between genetics and physiology. However, the application of deep learning to proteomics is currently hindered by a reliance on large-scale labeled datasets, rendering standard models ineffective for rare or novel diseases where patient samples are inherently scarce. RESULTS: Here, we present ProMeta, a meta-learning framework designed to enable robust disease modeling under extreme data restrictions. By integrating knowledge-guided pathway encoding with bi-level meta-optimization, ProMeta projects unstructured proteomic profiles into biologically interpretable functional tokens. This architecture allows the model to learn a global initialization containing transferable biological priors from biobank-scale data, facilitating rapid adaptation to novel tasks. Through comprehensive benchmark experiments, ProMeta consistently outperformed transfer learning and traditional machine learning baselines in both disease diagnosis and prediction tasks. In the most challenging 4-shot scenarios (utilizing only 2 cases and 2 controls), the model achieved robust generalization with an average AUROC of ∼0.69, representing a 24.6% relative improvement over the best-performing baseline methods. Mechanistic investigation revealed that ProMeta disentangles cases from controls in the latent space prior to task-specific adaptation, confirming the acquisition of universal biological rules rather than rote memorization. Furthermore, gradient-based interpretation identified disease-specific protein biomarkers and functional pathways consistent with known pathophysiology. Collectively, ProMeta overcomes the data-scarcity bottleneck in precision medicine, providing a scalable, interpretable framework for characterizing the full spectrum of human diseases, particularly for rare conditions lacking extensive clinical cohorts. AVAILABILITY AND IMPLEMENTATION: The source code of ProMeta is available at GitHub (https://github.com/lihan97/ProMeta).

Proteomics↗

PubMed2025

Antisense oligonucleotide depletion of CCDC146 is a broad-spectrum therapeutic strategy for ALS.

Amyotrophic lateral sclerosis (ALS) is a heritable and incurable disease defined by the degeneration of motor neurons (MNs), yet the genetics of ALS remain partially understood. Using a genomic deep learning-powered whole-genome analysis of 6,715 ALS patients, we identify four rare noncoding variants associated with patient survival, including chr7:76,009,472:C>T which is linked to a 70.6% reduction in survival. Genetic editing of this variant into iPSC-derived MNs increases CCDC146 expression and exacerbates ALS-specific phenotypes including TDP-43 mislocalization. We reveal that CCDC146 was located within the basal body of primary cilia in human MNs, and that cilia structure and function is impaired by CCDC146 overexpression but is restored by its depletion. Suppressing CCDC146 using an antisense oligonucleotide (ASO) completely rescues ALS-specific survival defects in neurons derived from both sporadic and familial patients, and it extends survival and reverses TDP-43 pathology in an aggressive ALS mouse model. Taken together, CCDC146 is a new modifier of ALS survival that acts via the primary cilia of MNs. ASO targeting of CCDC146 is a potential therapeutic approach for both sporadic and genetic forms of ALS, particularly because congenital absence of CCDC146 is well tolerated.

Journal Article↗

PubMed2025

Evaluation of a biomarker for amyotrophic lateral sclerosis derived from a hypomethylated DNA signature of human motor neurons.

Amyotrophic lateral sclerosis (ALS) lacks a specific biomarker, but is defined by relatively selective toxicity to motor neurons (MN). As others have highlighted, this offers an opportunity to develop a sensitive and specific biomarker based on detection of DNA released from dying MN within accessible biofluids. Here we have performed whole genome bisulfite sequencing (WGBS) of iPSC-derived MN from neurologically normal individuals. By comparing MN methylation with an atlas of tissue methylation we have derived a MN-specific signature of hypomethylated genomic regions, which accords with genes important for MN function. Through simulation we have optimised the selection of regions for biomarker detection in plasma and CSF cell-free DNA (cfDNA). However, we show that MN-derived DNA is not detectable via WGBS in plasma cfDNA. In support of our experimental finding, we show theoretically that the relative sparsity of lower MN sets a limit on the proportion of plasma cfDNA derived from MN which is below the threshold for detection via WGBS. Our findings are important for the ongoing development of ALS biomarkers. The MN-specific hypomethylated genomic regions we have derived could be usefully combined with more sensitive detection methods and perhaps with study of CSF instead of plasma. Indeed we demonstrate that neuronal-derived DNA is detectable in CSF. Our work is relevant for all diseases featuring death of rare cell-types.

Humans↗