Audemard, E. | Search bioRxiv

bioRxiv2023

SPAT: Surface Protein Annotation Tool

Given the particular attractivity of antibody-based immunotherapies, in vitro experimental approaches aiming to identify and quantify proteins directly located at the cell surface, such as the surfaceome, have been recently developed and improved. However, the "surface" enriched, yet noisy output obtained from available methods makes it challenging to accurately evaluate which proteins are more likely to be located at the surface of the plasma membrane and which are simple contaminants. To that purpose, we developed the in silico Surface Protein Annotation Tool (SPAT), which unifies established annotations to grade proteins according to the chance they have to be located at the cell surface. SPAT accuracy was tested using in-house acute myeloid leukemia data, as well as public datasets, and despite using publicly available annotations, showed good performances when compared to more complex surfaceome predictors. Given its simple input requirement, SPAT is easily usable for the annotation of any gene/protein lists. Its output, in addition to the "surface" score, provides additional annotations including a "secretion" flag, references to verified antibodies targeting annotated proteins, as well as expression data and protein levels in essential human organs, making it a user-friendly tool for the community.

bioinformatics↗

bioRxiv2018

Target variant detection in leukemia using unaligned RNA-Seq reads

Mutations identified in each Acute Myeloid Leukemia (AML) patients are useful for prognosis and to select targeted therapies. Detection of such mutations by the analysis of Next-Generation Sequencing (NGS) data requires a computationally intensive read mapping step and application of several variant calling methods. Targeted mutation identification drastically shifts the usual tradeoff between accuracy and performance by concentrating all computations over a small portion of sequence space. Here, we present km, an efficient approach leveraging k-mer decomposition of reads to identify targeted mutations. Our approach is versatile, as it can detect single-base mutations, several types of insertions and deletions, as well as fusions. We used two independent AML cohorts (The Cancer Genome Atlas and Leucegene), to show that mutation detection by km is fast, accurate and mainly limited by sequencing depth. Therefore, km allows to establish fast diagnostics from NGS data, and could be suitable for clinical applications.

bioinformatics↗