Transcriptomic Harmonization as the Way for Suppressing Cross-Platform Bias and Batch Effect

Abstract

(1) Background: Emergence of methods interrogating gene expression at high throughput gave birth to quantitative transcriptomics, but also posed a question of inter-comparison of expression profiles obtained using different equipment and protocols and/or in different series of experiments. Addressing this issue is challenging, because all of the above variables can dramatically influence gene expression signals and, therefore, cause a plethora of peculiar features in the transcriptomic profiles. Millions of transcriptomic profiles were obtained and deposited in public databases of which the usefulness is however strongly limited due to the inter-comparison issues; (2) Methods: Dozens of methods and software packages that can be generally classified as either flexible or predefined format harmonizers have been proposed, but none has become to the date the gold standard for unification of this type of Big Data; (3) Results: However, recent developments evidence that platform/protocol/batch bias can be efficiently reduced not only for the comparisons of limited transcriptomic datasets. Instead, instruments were proposed for transforming gene expression profiles into the universal, uniformly shaped format that can support multiple inter-comparisons for reasonable calculation costs. This forms a basement for universal indexing of all or most of all types of RNA sequencing and microarray hybridization profiles; (4) Conclusions: In this paper, we attempted to overview the landscape of modern approaches and methods in transcriptomic harmonization and focused on the practical aspects of their application.

A simplified machine learning model utilizing platelet-related genes for predicting poor prognosis in sepsis

“…Consequently, the data testing on this platform may yield lower AUC. To address these challenges, recent studies have proposed methods such as quantile normalization and cross platform normalization to mitigate cross-platform bias and batch effects ( 49 ) Future studies can employ these methods to further validate the conclusions of this study. Furthermore, we compared AUC of this model with the APACHE II and SOFA scores to evaluate its prognostic predictive effect.…”

Uniformly shaped harmonization combines human transcriptomic data from different platforms while retaining their biological properties and differential gene expression patterns

“… Shambhala-1/2 approach ( Borisov et al, 2019 ; 2022 ; Borisov and Buzdin, 2022 ) to uniformly shaped harmonization of gene expression data. Gene expression profiles for samples (1, … , N ), e.g., obtained using different experimental platforms are taken one by one, separately merged, and quantile-normalized with an auxiliary calibration dataset P .…”

“…We used Watermelon Multisection (WM) method to quantitatively assess the quality of clustering for the expression profiles under analysis according to ( Zolotovskaia M. A. et al, 2020 ; Borisov and Buzdin, 2022 ; Borisov et al, 2022 ). This method returns a specific metric for the assessment of an entropy-based quality of clustering on dendrograms according to known predefined classes.…”

“…We then introduced a new type of uniformly shaped cross-platform harmonizers that use several mathematical transforms ( Borisov et al, 2019 ; 2022 ; Borisov and Buzdin, 2022 ). The first version of such software, Shambhala-1 ( Borisov et al, 2019 ) used uniformly shaped harmonization that employed the piecewise-linear gene expression transformation method XPN ( Shabalin et al, 2008 ; Rudy and Valafar, 2011 ).…”

A General Primer for Data Harmonization

“…Meanwhile, other researchers have proposed statistical techniques to account for measurement errors, from simpler strategies like pre-processing 122 or outlier detection methods 123 to more complex model-based techniques, like e.g. linear or deep learning models, which tend to be quite specific to different fields and datasets 123 – 127 .…”