Single sample scoring of molecular phenotypes

Abstract

BackgroundGene set scoring provides a useful approach for quantifying concordance between sample transcriptomes and selected molecular signatures. Most methods use information from all samples to score an individual sample, leading to unstable scores in small data sets and introducing biases from sample composition (e.g. varying numbers of samples for different cancer subtypes). To address these issues, we have developed a truly single sample scoring method, and associated R/Bioconductor package singscore (https://bioconductor.org/packages/singscore).ResultsWe use multiple cancer data sets to compare singscore against widely-used methods, including GSVA, z-score, PLAGE, and ssGSEA. Our approach does not depend upon background samples and scores are thus stable regardless of the composition and number of samples being scored. In contrast, scores obtained by GSVA, z-score, PLAGE and ssGSEA can be unstable when less data are available (NS < 25). The singscore method performs as well as the best performing methods in terms of power, recall, false positive rate and computational time, and provides consistently high and balanced performance across all these criteria. To enhance the impact and utility of our method, we have also included a set of functions implementing visual analysis and diagnostics to support the exploration of molecular phenotypes in single samples and across populations of data.ConclusionsThe singscore method described here functions independent of sample composition in gene expression data and thus it provides stable scores, which are particularly useful for small data sets or data integration. Singscore performs well across all performance criteria, and includes a suite of powerful visualization functions to assist in the interpretation of results. This method performs as well as or better than other scoring approaches in terms of its power to distinguish samples with distinct biology and its ability to call true differential gene sets between two conditions. These scores can be used for dimensional reduction of transcriptomic data and the phenotypic landscapes obtained by scoring samples against multiple molecular signatures may provide insights for sample stratification.Electronic supplementary materialThe online version of this article (10.1186/s12859-018-2435-4) contains supplementary material, which is available to authorized users.

Stable gene expression for normalisation and single-sample scoring

“…Additionally, we showed that context/tissue-specific processes such as epithelial-mesenchymal transition were enriched with highly variable genes, as expected given the diverse changes associated with this phenotypic program (43). We used singscore (35) with stability ranks to enable this analysis, demonstrating that singscore can be used to assess enrichment using any ranked data.…”

“…Our aim was to evaluate enrichment accounting for the prioritisation of stability in our list. We adapted singscore (35) to achieve this and computed stability scores for gene sets derived from gene ontologies and KEGG pathways. Gene sets associated with RNA processing (GO:0006396, GO:0008380), the spliceosome complex (GO:0005681), mRNA metabolic process…”

“…We previously developed a method, singscore, to score individual samples against gene set signatures using transcriptomic data and showed that these scores can assist in assessing the molecular phenotype of tissues and cell lines (35). Though the method has been applied in diverse scenarios in an exploratory context (25,35,36,42,43), the potential for clinical translation is limited by a requirement for transcriptome-wide measurements.…”

“…Scores using both up-and down-regulated genes were centred around 0 by subtracting the median score (0.5) from them resulting, thus ensuring score were in the range [-0.5, 0.5]. Additionally, signatures composed of both up-and down-regulated genes with unknown direction can be scored using an approach similar to singscore (35).…”

Reprogramming of serine metabolism is an actionable vulnerability in FLT3-ITD driven acute myeloid leukaemia

“…E. Analysis of a genetic signature encompassing genes involved in de novo serine synthesis, one-carbon/folate cycle, and rate-limiting de novo purine biosynthesis pathways (SSP-OC-Purine Signature). mRNA counts from the LAML-TCGA dataset were subset on FLT3-mutant (n = 41) and FLT3 wild-type (n = 100) samples, and the degree of correlation of the signature against FLT3 mRNA expression in each subset was performed using the singscore scoring method (45).…”