Miller, D. B. | Search bioRxiv

bioRxiv2021

Benchmarking 50 classification algorithms on 50 gene-expression datasets

By classifying patients into subgroups, clinicians can provide more effective care than using a uniform approach for all patients. Such subgroups might include patients with a particular disease subtype, patients with a good (or poor) prognosis, or patients most (or least) likely to respond to a particular therapy. Diverse types of biomarkers have been proposed for assigning patients to subgroups. For example, DNA variants in tumors show promise as biomarkers; however, tumors exhibit considerable genomic heterogeneity. As an alternative, transcriptomic measurements reflect the downstream effects of genomic and epigenomic variations. However, high-throughput technologies generate thousands of measurements per patient, and complex dependencies exist among genes, so it may be infeasible to classify patients using traditional statistical models. Machine-learning classification algorithms can help with this problem. However, hundreds of classification algorithms exist--and most support diverse hyperparameters--so it is difficult for researchers to know which are optimal for gene-expression biomarkers. We performed a benchmark comparison, applying 50 classification algorithms to 50 gene-expression datasets (143 class variables). We evaluated algorithms that represent diverse machine-learning methodologies and have been implemented in general-purpose, open-source, machine-learning libraries. When available, we combined clinical predictors with gene-expression data. Additionally, we evaluated the effects of performing hyperparameter optimization and feature selection in nested cross-validation folds. Kernel- and ensemble-based algorithms consistently outperformed other types of classification algorithms; however, even the top-performing algorithms performed poorly in some cases. Hyperparameter optimization and feature selection typically improved predictive performance, and univariate feature-selection algorithms outperformed more sophisticated methods. Together, our findings illustrate that algorithm performance varies considerably when other factors are held constant and thus that algorithm selection is a critical step in biomarker studies. Author Summary

bioinformatics↗

bioRxiv2020

Cingulo-Opercular Control Network Supports Disused Motor Circuits in Standby Mode

Whole-brain resting-state functional MRI (rs-fMRI) during two weeks of limb constraint revealed that disused motor regions became more strongly connected to the cingulo-opercular network (CON), an executive control network that includes regions of the dorsal anterior cingulate cortex (dACC) and insula (1). Disuse-driven increases in functional connectivity (FC) were specific to the CON and somatomotor networks and did not involve any other networks, such as the salience, frontoparietal, or default mode networks. Censoring and modeling analyses showed that FC increases during casting were mediated by large, spontaneous activity pulses that appeared in the disused motor regions and CON control regions. During limb constraint, disused motor circuits appear to enter a standby mode characterized by spontaneous activity pulses and strengthened connectivity to CON executive control regions. SignificanceMany studies have examined plasticity in the primary somatosensory and motor cortex during disuse, but little is known about how disuse impacts the brain outside of primary cortical areas. We leveraged the whole-brain coverage of resting-state functional MRI (rs-fMRI) to discover that disuse drives plasticity of distant executive control regions in the cingulo-opercular network (CON). Two complementary analyses, pulse censoring and pulse addition, demonstrated that increased functional connectivity between the CON and disused motor regions was driven by large, spontaneous pulses of activity in the CON and disused motor regions. These results point to a previously unknown role for the CON in supporting motor plasticity and reveal spontaneous activity pulses as a novel mechanism for reorganizing the brains functional connections.

neuroscience↗