A novel batch-effect correction method for scRNA-seq data based on Adversarial Information Factorization.

Single-cell RNA sequencing (scRNA-seq) technology produces an unprecedented resolution at the level of a unique cell, raising great hopes in medicine. Nevertheless, scRNA-seq data suffer from high variations due to the experimental conditions, called batch effects, preventing any aggregated downstream analysis. Adversarial Information Factorization provides a robust batch-effect correction method that does not rely on prior knowledge of the cell types nor a specific normalization strategy while being adapted to any downstream analysis task. It compares to and even outperforms state-of-the-art methods in several scenarios: low signal-to-noise ratio, batch-specific cell types with few cells, and a multi-batches dataset with imbalanced batches and batch-specific cell types. Moreover, it best preserves the relative gene expression between cell types, yielding superior differential expression analysis results. Finally, in a more complex setting of a Leukemia cohort, our method preserved most of the underlying biological information for each patient while aligning the batches, improving the clustering metrics in the aggregated dataset.

Molecular International Prognostic Scoring System for Myelodysplastic Syndromes.

BACKGROUND: Risk stratification and therapeutic decision-making for myelodysplastic syndromes (MDS) are based on the International Prognostic Scoring System­Revised (IPSS-R), which considers hematologic parameters and cytogenetic abnormalities. Somatic gene mutations are not yet used in the risk stratification of patients with MDS. METHODS: To develop a clinical-molecular prognostic model (IPSS-Molecular [IPSS-M]), pretreatment diagnostic or peridiagnostic samples from 2957 patients with MDS were profiled for mutations in 152 genes. Clinical and molecular variables were evaluated for associations with leukemia-free survival, leukemic transformation, and overall survival. Feature selection was applied to determine the set of independent IPSS-M prognostic variables. The relative weights of the selected variables were estimated using a robust Cox multivariable model adjusted for confounders. The IPSS-M was validated in an external cohort of 754 Japanese patients with MDS. RESULTS: We mapped at least one oncogenic genomic alteration in 94% of patients with MDS. Multivariable analysis identified TP53multihit, FLT3 mutations, and MLLPTD as top genetic predictors of adverse outcomes. Conversely, SF3B1 mutations were associated with favorable outcomes, but this was modulated by patterns of comutation. Using hematologic parameters, cytogenetic abnormalities, and somatic mutations of 31 genes, the IPSS-M resulted in a unique risk score for individual patients. We further derived six IPSS-M risk categories with prognostic differences. Compared with the IPSS-R, the IPSS-M improved prognostic discrimination across all clinical end points and restratified 46% of patients. The IPSS-M was applicable in primary and secondary/therapy-related MDS. To simplify clinical use of the IPSS-M, we developed an open-access Web calculator that accounts for missing values. CONCLUSIONS: Combining genomic profiling with hematologic and cytogenetic parameters, the IPSS-M improves the risk stratification of patients with MDS and represents a valuable tool for clinical decision-making. (Funded by Celgene Corporation through the MDS Foundation, the Josie Robertson Investigators Program, the Edward P. Evans Foundation, the Projects of National Relevance of the Italian Ministry of University and Research, Associazione Italiana per la Ricerca sul Cancro, the Japan Agency for Medical Research and Development, Cancer Research UK, the Austrian Science Fund, the MEXT [Japanese Ministry of Education, Culture, Sports, Science and Technology] Program for Promoting Research on the Supercomputer Fugaku, the Japan Society for the Promotion of Science, the Taiwan Department of Health, and Celgene Corporation through the MDS Foundation.)