Identification of novel protein biomarkers from the blood and urine for the early diagnosis of bladder cancer via proximity extension analysis.

BACKGROUND: Bladder cancer (BC) is a very common urinary tract malignancy that has a high incidence and lethality. In this study, we identified BC biomarkers and described a new noninvasive detection method using serum and urine samples for the early detection of BC. METHODS: Serum and urine samples were retrospectively collected from patients with BC (n = 99) and healthy controls (HC) (n = 50), and the expression levels of 92 inflammation-related proteins were examined via the proximity extension analysis (PEA) technique. Differential protein expression was then evaluated by univariate analysis (p < 0.05). The expression of the selected potential marker was further verified in BC and adjacent tissues by immunohistochemistry (IHC) and single-cell sequencing. A model was constructed to differentiate BC from HC by LASSO regression and compared to the detection capability of FISH. RESULTS: The univariate analysis revealed significant differences in the expression levels of 40 proteins in the serum (p < 0.05) and 17 proteins in the urine (p < 0.05) between BC patients and HC. Six proteins (AREG, RET, WFDC2, FGFBP1, ESM-1, and PVRL4) were selected as potential BC biomarkers, and their expression was evaluated at the protein and transcriptome levels by IHC and single-cell sequencing, respectively. A diagnostic model (a signature) consisting of 14 protein markers (11 in serum and three in urine) was also established using LASSO regression to distinguish between BC patients and HC (area under the curve = 0.91, PPV = 0.91, sensitivity = 0.87, and specificity = 0.82). Our model showed better diagnostic efficacy than FISH, especially for early-stage, small, and low-grade BC. CONCLUSION: Using the PEA method, we identified a panel of potential protein markers in the serum and urine of BC patients. These proteins are associated with the development of BC. A total of 14 of these proteins can be used to detect early-stage, small, low-grade BC. Thus, these markers are promising for clinical translation to improve the prognosis of BC patients.

Ensemble Learning with Supervised Methods Based on Large-Scale Protein Language Models for Protein Mutation Effects Prediction.

Machine learning has been increasingly utilized in the field of protein engineering, and research directed at predicting the effects of protein mutations has attracted increasing attention. Among them, so far, the best results have been achieved by related methods based on protein language models, which are trained on a large number of unlabeled protein sequences to capture the generally hidden evolutionary rules in protein sequences, and are therefore able to predict their fitness from protein sequences. Although numerous similar models and methods have been successfully employed in practical protein engineering processes, the majority of the studies have been limited to how to construct more complex language models to capture richer protein sequence feature information and utilize this feature information for unsupervised protein fitness prediction. There remains considerable untapped potential in these developed models, such as whether the prediction performance can be further improved by integrating different models to further improve the accuracy of prediction. Furthermore, how to utilize large-scale models for prediction methods of mutational effects on quantifiable properties of proteins due to the nonlinear relationship between protein fitness and the quantification of specific functionalities has yet to be explored thoroughly. In this study, we propose an ensemble learning approach for predicting mutational effects of proteins integrating protein sequence features extracted from multiple large protein language models, as well as evolutionarily coupled features extracted in homologous sequences, while comparing the differences between linear regression and deep learning models in mapping these features to quantifiable functional changes. We tested our approach on a dataset of 17 protein deep mutation scans and indicated that the integrated approach together with linear regression enables the models to have higher prediction accuracy and generalization. Moreover, we further illustrated the reliability of the integrated approach by exploring the differences in the predictive performance of the models across species and protein sequence lengths, as well as by visualizing clustering of ensemble and non-ensemble features.