No abstract

The Negative Binomial distribution (NBD) is used for modeling many types of count data, including gene expression counts obtained by RNA sequencing technologies (RNA-Seq). Finding outliers in this type of data has been shown in recent research to help in identifying rare genetic disorders in humans. Existing Bayesian approaches to detecting outliers in data following the NBD are either computationally inefficient or too general and hence do not leverage the NBD's specificities in an optimal way. We present a novel Bayesian outlier score for data following the NBD, relying on recent advances in the inference of its dispersion parameter through a special method of Gibbs sampling. The novel Bayesian model on which our score is based -OutPyRX (Outlier detection in Python for RNA-Seq, eXtended version) -improves the model of its predecessor OutPyR by introducing novel parameters that are derived from OutPyR's. These novel parameters allow more than 6 times faster convergence of the novel outlier score compared to OutPyR's while having a negligible computational impact on the Gibbs sampling procedure. We show that, in terms of area under precision-recall curve (AUC) values, the novel score outcompetes existing scores on 21 out of 24 datasets that we derived from 4 real datasets by injecting artificial outliers. However, OutPyRX does not perform confounder control which is required for some datasets containing biological outliers. The model is general and can be applied to other similar count data. The code for our model is available at https://github.com/esalkovic/outpyrx.

Motivation Finding outliers in RNA Sequencing (RNA-Seq) gene expression (GE) can help in identifying genes that are aberrant and cause Mendelian disorders. Recently developed models for this task rely on modeling RNA-Seq GE data using the Negative Binomial distribution (NBD). However, some of those models either rely on procedures for inferring NBD’s parameters in a non-biased way that are computationally demanding, and thus make confounder control challenging, while others rely on less computationally demanding but biased procedures and convoluted confounder control approaches that hinder interpretability. Results In this paper we present OutSingle (Outlier detection using Singular Value Decomposition), an almost instantaneous way of detecting outliers in RNA-Seq GE data. It uses a simple log-normal approach for count modeling. For confounder control it uses the recently discovered Optimal Hard Threshold (OHT) method for noise detection, which itself is based on Singular Value Decomposition (SVD). Due to its SVD/OHT utilization, OutSingle’s model is straightforward to understand and interpret. We then show that our novel method, when used on RNA-Seq GE data with real biological outliers masked by confounders, outcompetes the previous state-of-the art model based on an ad-hoc denoising autoencoder (AE). Additionally, OutSingle can be used to inject artificial outliers masked by confounders, which is difficult to achieve with previous approaches. We describe a way of using OutSingle for outlier injection and proceed to show how OutSingle outperforms its competition on 16 out of 18 datasets that were generated from 3 real datasets using OutSingle’s injection procedure with different outlier types and magnitudes. Our methods are applicable to other types of similar problems involving finding outliers in matrices under the presence of confounders. Availability The code for OutSingle is available at https://github.com/esalkovic/outsingle Supplementary information Supplementary data are available at Bioinformatics online.