Accelerating genomic workflows using NVIDIA Parabricks.

BACKGROUND: As genome sequencing becomes better integrated into scientific research, government policy, and personalized medicine, the primary challenge for researchers is shifting from generating raw data to analyzing these vast datasets. Although much work has been done to reduce compute times using various configurations of traditional CPU computing infrastructures, Graphics Processing Units (GPUs) offer opportunities to accelerate genomic workflows by orders of magnitude. Here we benchmark one GPU-accelerated software suite called NVIDIA Parabricks on Amazon Web Services (AWS), Google Cloud Platform (GCP), and an NVIDIA DGX cluster. We benchmarked six variant calling pipelines, including two germline callers (HaplotypeCaller and DeepVariant) and four somatic callers (Mutect2, Muse, LoFreq, SomaticSniper). RESULTS: We achieved up to 65 × acceleration with germline variant callers, bringing HaplotypeCaller runtimes down from 36 h to 33 min on AWS, 35 min on GCP, and 24 min on the NVIDIA DGX. Somatic callers exhibited more variation between the number of GPUs and computing platforms. On cloud platforms, GPU-accelerated germline callers resulted in cost savings compared with CPU runs, whereas some somatic callers were more expensive than CPU runs because their GPU acceleration was not sufficient to overcome the increased GPU cost. CONCLUSIONS: Germline variant callers scaled well with the number of GPUs across platforms, whereas somatic variant callers exhibited more variation in the number of GPUs with the fastest runtimes, suggesting that, at least with the version of Parabricks used here, these workflows are less GPU optimized and require benchmarking on the platform of choice before being deployed at production scales. Our study demonstrates that GPUs can be used to greatly accelerate genomic workflows, thus bringing closer to grasp urgent societal advances in the areas of biosurveillance and personalized medicine.

ER, Mitochondria, and ISR Regulation by mt-HSP70 and ATF5 upon Procollagen Misfolding in Osteoblasts.

Cellular response to protein misfolding underlies multiple diseases. Collagens are the most abundant vertebrate proteins, yet little is known about cellular response to misfolding of their procollagen precursors. Osteoblasts (OBs)-the cells that make bone-produce so much procollagen that it accounts for up to 40% of mRNAs in the cell, which is why bone bears the brunt of mutations causing procollagen misfolding in osteogenesis imperfecta (OI). The present study of a G610C mouse model of OI by multiple transcriptomic techniques provides first solid clues to how OBs respond to misfolded procollagen accumulation in the endoplasmic reticulum (ER) and how this response affects OB function. Surprisingly, misfolded procollagen escapes the quality control in the ER lumen and indirectly triggers the integrated stress response (ISR) through other cell compartments. In G610C OBs, the ISR is regulated by mitochondrial HSP70 (mt-HSP70) and ATF5 instead of their BIP and ATF4 paralogues, which normally activate and regulate ISR to secretory protein misfolding in the ER. The involvement of mt-HSP70 and ATF5 together with other transcriptomic findings suggest that mitochondria might initiate the ISR upon disruption of ER-mitochondria connections or might respond to the ISR activated by a yet unknown sensor.

Noncanonical ER-Golgi trafficking and autophagy of endogenous procollagen in osteoblasts.

Secretion and quality control of large extracellular matrix proteins remain poorly understood and debated, particularly transport intermediates delivering folded proteins from the ER to Golgi and misfolded ones to lysosomes. Discrepancies between different studies are related to utilization of exogenous cargo, off-target effects of experimental conditions and cell manipulation, and identification of transport intermediates without tracing their origin and destination. To address these issues, here we imaged secretory and degradative trafficking of type I procollagen in live MC3T3 osteoblasts by replacing a region encoding N-propeptide in endogenous Col1a2 gDNA with GFP cDNA. We selected clones that produced the resulting fluorescent procollagen yet had normal expression of key osteoblast and ER/cell stress genes, normal procollagen folding, and normal deposition and mineralization of extracellular matrix. Live-cell imaging of these clones revealed ARF1-dependent transport intermediates, which had no COPII coat and delivered procollagen from ER exit sites (ERESs) to Golgi without stopping at ER-Golgi intermediate compartment (ERGIC). It also confirmed ERES microautophagy, i.e., lysosomes engulfing ERESs containing misfolded procollagen. Beyond validating these trafficking models for endogenous procollagen, we uncovered a probable cause of noncanonical cell stress response to procollagen misfolding. Recognized and retained only at ERESs, misfolded procollagen does not directly activate the canonical UPR, yet it disrupts the ER lumen by blocking normal secretory export from the ER.

Mechanisms of procollagen and HSP47 sorting during ER-to-Golgi trafficking.

Efficient quality control and export of procollagen from the cell is crucial for extracellular matrix homeostasis, yet it is still incompletely understood. One of the debated questions is the role of a collagen-specific ER chaperone HSP47 in these processes. Most ER chaperones preferentially bind to unfolded polypeptide chains, enabling selective export of natively folded proteins from the ER after chaperone release. In contrast, HSP47 preferentially binds to the natively folded procollagen and is believed to be released only in the ER-Golgi intermediate compartment (ERGIC) or cis-Golgi. HSP47 colocalization with procollagen in punctate structures observed by immunofluorescence imaging of fixed cells has thus been interpreted as evidence for HSP47 export from the ER together with procollagen in transport vesicles destined for ERGIC or Golgi. To understand the mechanism of this co-trafficking and its physiological significance, we imaged the dynamics of fluorescently tagged type I procollagen and HSP47 punctate structures in live MC3T3 murine osteoblasts with up to 120 nm spatial and 500 ms time resolution. Contrary to the prevailing model, we discovered that most bona fide carriers delivering procollagen from ER exit sites (ERESs) to Golgi contained no HSP47, unless the RDEL signal for ER retention in HSP47 was deleted or mutated. These transport intermediates exhibited characteristic rapid, directional motion along microtubules, while puncta with colocalized HSP47 and procollagen similar to the ones described before had only limited, stochastic motion. Live cell imaging and fluorescence recovery after photobleaching revealed that the latter puncta (including the ones induced by ARF1 inhibition) were dilated regions of ER lumen, ERESs, or autophagic structures surrounded by lysosomal membranes. Procollagen was colocalized with HSP47 and ERGIC53 at ERESs. It was colocalized with ERGIC53 but not HSP47 in Golgi-bound transport intermediates. Our results suggest that procollagen and HSP47 sorting occurs at ERES before procollagen is exported from the ER in Golgi-bound transport intermediates, providing new insights into mechanisms of procollagen trafficking.