LinDeconSeq

"LinDeconSeq" is a hybrid computational tool for deconvolution of cellular fractions from gene expression data of bulk samples. It focuses particularly on the identification of cell type-specific genes (markers) across multiple conditions. This tool addresses the challenge of accurately representing gene expression specificity in complex biological samples, where traditional pairwise comparisons may fall short.

Key Features and Functionalities:

- Specificity Scoring and Mutual Linearity: LinDeconSeq uniquely identifies marker genes by combining specificity scoring with mutual linearity strategies, allowing for the identification of markers across any number of cell types, enhancing the tool's versatility and applicability.

- Weighted Robust Linear Regression: In its second stage, LinDeconSeq predicts cellular fractions of bulk samples using weighted robust linear regression based on previously identified marker genes. This approach ensures accurate and reliable deconvolution of cellular compositions within complex samples.

- Improved Accuracy and Reproducibility: Benchmarking against multiple publicly available datasets has demonstrated that the marker genes identified by LinDeconSeq offer better accuracy and reproducibility than other methods like MGFM and RNentropy.

- Clinical Relevance: LinDeconSeq has been applied to diagnose acute myeloid leukemia (AML), successfully identifying distinct cellular fractions associated with the disease. The tool's ability to differentiate between patient subgroups based on cellular composition and its correlation with prognosis and mutation patterns underscores its potential clinical utility.