CellHeap: A scRNA-seq workflow for large-scalebioinformatics data analysis
AbstractO_ST_ABSBackgroundC_ST_ABSSeveral hundred terabytes of single-cell RNA-seq (scRNA-seq) data are available in public repositories. These data refer to various research projects, from microbial population cells to multiple tissues, involving patients with a myriad of diseases and comorbidities. An increase to several Petabytes of scRNA-seq data available in public repositories is a realistic prediction for coming years. Therefore, thoughtful analysis of these data requires large-scale computing infrastructures and software systems optimized for such platforms to generate correct and reliable biological knowledge. ResultsThis paper presents CellHeap, a flexible, portable, and robust platform for analyzing large scRNA-seq datasets, with quality control throughout the execution steps, and deployable on platforms that support large-scale data, such as supercomputers or clouds. As a case study, we designed a workflow to study particular modulations of Fc receptors, considering mild and severe cases of COVID-19. This workflow, deployed in the Brazilian Santos Dumont supercomputer, processed dozens of Terabytes of COVID-19 scRNA-seq raw data. Our results show that most of the workflow total execution time is spent in its initial phases and that there is great potential for a parallel solution to speed up scRNA-seq data analysis significantly. Thus, this workflow includes an efficient solution to use parallel computational resources, improving total execution time. Our case study showed increased Fc receptors transcription in macrophages of patients with severe COVID-19 symptoms, especially FCGR1A, FCGR2A, and FCGR3A. Furthermore, diverse molecules associated with their signaling pathways were upregulated in severe cases, possibly associated with the prominent inflammatory response observed. ConclusionFrom the CellHeap platform, different workflows capable of analyzing large scRNA-seq datasets can be generated. Our case study, a workflow designed to study particular modulations of Fc receptors, considering mild and severe cases of COVID-19, deployed on the Brazilian supercomputer Santos Dumont, had a substantial reduction in total execution time when jobs are triggered simultaneously using the parallelization strategy described in this manuscript. Regarding biological results, our case study identified specific modulations comparing healthy individuals with COVID-19 patients with mild or severe symptoms, revealing an upregulation of several inflammatory pathways and an increase in the transcription of Fc receptors in severe cases.