Peroxisome biogenesis initiated by protein phase separation.

Peroxisomes are organelles that carry out ß-oxidation of fatty acids and amino acids. Both rare and prevalent diseases are caused by their dysfunction1. Among disease-causing variant genes are those required for protein transport into peroxisomes. The peroxisomal protein import machinery, which also shares similarities with chloroplasts2, is unique in transporting folded and large, up to 10 nm in diameter, protein complexes into peroxisomes3. Current models postulate a large pore formed by transmembrane proteins4; however, so far, no pore structure has been observed. In the budding yeast Saccharomyces cerevisiae, the minimum transport machinery includes the membrane proteins Pex13 and Pex14 and the cargo-protein-binding transport receptor, Pex5. Here we show that Pex13 undergoes liquid-liquid phase separation (LLPS) with Pex5-cargo. Intrinsically disordered regions in Pex13 and Pex5 resemble those found in nuclear pore complex proteins. Peroxisomal protein import depends on both the number and pattern of aromatic residues in these intrinsically disordered regions, consistent with their roles as 'stickers' in associative polymer models of LLPS5,6. Finally, imaging fluorescence cross-correlation spectroscopy shows that cargo import correlates with transient focusing of GFP-Pex13 and GFP-Pex14 on the peroxisome membrane. Pex13 and Pex14 form foci in distinct time frames, suggesting that they may form channels at different saturating concentrations of Pex5-cargo. Our findings lead us to suggest a model in which LLPS of Pex5-cargo with Pex13 and Pex14 results in transient protein transport channels7.

Enzyme-Instructed Photoactivatable Supramolecular Antigens on Cancer Cell Membranes for Precision-Controlled T-Cell-Based Cancer Immunotherapy.

Cancer immunotherapies based on cytotoxic CD8+ T lymphocytes (CTLs) are highly promising for cancer treatment. The specific interaction between T-cell receptors and peptide-MHC-I complexes (pMHC-I) on cancer cell membranes critically determines their therapeutic outcomes. However, the lack of appropriate endogenous antigens for MHC-I presentation disables tumor recognition by CTLs. By devising three antigen-loaded self-assembling peptides of pY-K(Ag)-ERGD, pY-K(Ag)-E, and Y-K(Ag)-ERGD to noncovalently generate light-activatable supramolecular antigens at tumor sites in different manners, we report pY-K(Ag)-ERGD as a promising candidate to endow tumor cells with pMHC-I targets on demand. Specifically, pY-K(Ag)-ERGD first generates low-antigenic supramolecular antigens on cancer cell membranes, and a successive light pulse allows antigen payloads to efficiently release from the supramolecular scaffold, directly producing antigenic pMHC-I. Intravenous administration of pY-K(Ag)-ERGD enables light-controlled tumor inhibition when combined with adoptively transferred antigen-specific CTLs. Our strategy is feasible for broadening tumor antigen repertoires for T-cell immunotherapies and advancing precision-controlled T-cell immunotherapies.

Switchable Pickering Emulsions Stabilized via Synergistic Nanoparticles-Superamphiphiles Effects and Rapid Response to CO2/N2.

A CO2/N2-responsive emulsion provides milder reaction conditions, nontoxicity, and economic feasibility compared to other switchable surfactants. In this study, CO2/N2-responsive pickering emulsions were fabricated by using a compounded dispersion containing SiO2 nanoparticles (NPs) and superamphiphiles as the emulsifying agents. The synergistic effects of the SiO2 NPs and superamphiphiles significantly stabilized the emulsion at all of the tested concentrations and prevented complete phase separation of oil and water. The electrostatic interaction between the SiO2 NPs and superamphiphiles was disrupted after bubbling with CO2 for 30 s, resulting in the breaking of the emulsion. However, the dispersion recovered its interfacial activity after the introduction of N2 and again emulsified the emulsion. This reversible switching behavior was validated through three consecutive cycles of bubbling CO2/N2. The protonation and deprotonation of the SiO2 NPs and superamphiphiles in response to CO2/N2 facilitated reversible assembly and disassembly, which enabled the switching of the emulsions between inactive and active forms. The novel highly stable Pickering emulsions demonstrated rapid demulsification and emulsification in response to CO2/N2 and are promising for a wide range of applications.

Experimental study of a thin thermally grown oxide layer in thermal barrier coatings based on the SWT-BP algorithm and terahertz technology.

As a promising nondestructive testing (NDT) technique with a very adaptive physical modeling of wave transmission process, terahertz technology is used for the detection and characterization of nonpolar materials and the evaluation of layered and/or defective structures. THz-TDS can also be used to perform spectroscopic analysis and detect structural defects in thermal barrier coatings (TBCs) of aero-engines. Although it is generally difficult to measure the structure of the thin oxide layer of the thermal barrier coatings whose thickness is generally lower than 30 µm (the current axial resolution of the THz-TDS cannot exceed 30 µm). We were able to complete the detection of the oxide layer within 1-29 µm through simulation by using the SWT-BP algorithm. In this study, the analysis was performed on real-world samples, the fitting degree of the SWT-BP algorithm reached 0.77, and the minimum prediction error was less than 0.1 µm. The paper also put forward some improvement measures about the experimental results.

HSFAS mediates fibroblast proliferation, migration, trans-differentiation and apoptosis in hypertrophic scars via interacting with ADAMTS8.

Hypertrophic scar (HS) is one of the most common sequelae of patients, especially after burns and trauma. The roles of regulatory long noncoding RNAs (lncRNAs) in mediating HS remain underexplored. Human hypertrophic scar-derived fibroblasts (HSFBs) have been shown to exert more potent promoting effects on extracellular matrix (ECM) accumulation than normal skin-derived fibroblasts (NSFBs) and are associated with enhanced HS formation. The purpose of this study is to search for lncRNAs enriched in HSFBs and investigate their roles and mechanisms. LncRNA MSTRG.59347.16 is one of the most highly expressed lncRNAs in HS detected by lncRNA-seq and qRT-PCR and named as hypertrophic scar fibroblast-associated lncRNA (HSFAS). HSFAS overexpression significantly induces fibroblast proliferation, migration, and myofibroblast trans-differentiation and inhibits apoptosis in HSFBs, while knockdown of HSFAS results in augmented apoptosis and attenuated proliferation, migration, and myofibroblast trans-differentiation of HSFBs. Mechanistically, HSFAS suppresses the expression of A disintegrin and metalloproteinase with thrombospondin motifs 8 (ADAMTS8). ADAMTS8 knockdown rescues downregulated HSFAS-mediated fibroblast proliferation, migration, myofibroblast trans-differentiation and apoptosis. Thus, our findings uncover a previously unknown lncRNA-dependent regulatory pathway for fibroblast function. Targeted intervention in the HSFAS-ADAMTS8 pathway is a potential therapy for HS.

In Situ Construction of Ferrocene-Containing Membrane-Bound Nanofibers for the Redox Control of Cancer Cell Death and Cancer Therapy.

Precise manipulation of cancer cell death by harnessing reactive oxygen species (ROS) is a promising strategy to defeat malignant tumors. However, it is quite difficult to produce active ROS with spatial precision and regulate their biological outcomes. We succeed here in selectively generating short-lived and lipid-reactive hydroxyl radicals (•OH) adjacent to cancer cell membranes, successively eliciting lipid peroxidation and ferroptosis. DiFc-K-pY, a phosphorylated self-assembling precursor that consists of two branched Fc moieties and interacts specifically with epidermal growth factor receptor, can in situ produce membrane-bound nanofibers and enrich ferrocene moieties on cancer cell membranes in response to alkaline phosphatase. Within the acidic tumor microenvironment, DiFc-K-pY nanofibers efficiently convert tumoral H2O2 to active •OH around the target cell membranes via Fenton-like reactions, leading to lipid peroxidation and ferroptosis with good cellular selectivity. Our strategy successfully prevents tumor progression with acceptable biocompatibility through intratumoral administration.

Copper-Induced Supramolecular Peptide Assemblies for Multi-pathway Cell Death and Tumor Inhibition.

Although self-assembly has emerged as an effective tool for fabricating biomaterials, achieving precise control over the morphologies and functionalities of the resultant assemblies remains an ongoing challenge. Inspired by the copper peptide naturally present in human plasma, in this study, we designed a synthetic precursor, FcGH. FcGH can self-assemble via two distinct pathways: spontaneous and Cu2+-induced. These two assembly pathways enabled the formation of assemblies with tunable morphologies by adjusting the amount of added Cu2+. We found that the nanoparticles formed by Cu2+-induced self-assembly exhibited a significantly higher cellular uptake efficiency than the wormlike fibers formed spontaneously. Moreover, this Cu2+-induced assembly process occurred spontaneously at a 1:1 molar ratio of Cu2+ to FcGH, avoiding the excessive use of Cu²âº and a tedious preparation procedure. By co-assembling with FcGH-conjugated 10-hydroxycamptothecin (HCPT), Cu2+-induced supramolecular nanodrugs elicited multiple cell death modalities in cancer cells with elevated immunogenicity, enhancing the therapeutic effect compared to free HCPT. This study highlights Cu2+-induced self-assembly as an efficient tool for directing the assembly of nanodrugs and for synergistic tumor therapy.

Micelles Cascade Assembly to Tandem Porous Catalyst for Waste Plastics Upcycling.

Catalytic upcycling of polyolefins into high-value chemicals represents the direction in end-of-life plastics valorization, but poses great challenges. Here, we report the synthesis of a tandem porous catalyst via a micelle cascade assembly strategy for selectively catalytic cracking of polyethylene into olefins at a low temperature. A hierarchically porous silica layer from mesopore to macropore is constructed on the surface of microporous ZSM-5 nanosheets through cascade assembly of dynamic micelles. The outer macropore arrays can adsorb bulky polyolefins quickly by the capillary and hydrophobic effects, enhancing the diffusion and access to active sites. The middle mesopores present a nanoconfinement space, pre-cracking polyolefins into intermediates by weak acid sites, which then transport into zeolites micropores for further cracking by strong Brønsted acid sites. The hierarchically porous and acidic structures, mimicking biomimetic protease catalytic clefts, ideally match the tandem cracking steps of polyolefins, thus suppressing coke formation and facilitating product escape. As a result, light hydrocarbons (C1-C7) are produced with a yield of 443 mmol gZSM-5 -1, where 74.3 % of them are C3-C6 olefins, much superior to ZSM-5 and porous silica catalysts. This tandem porous catalyst exemplifies a superstructure design of catalytic cracking catalysts for industrial and economical upcycling of plastic wastes.

Enzyme-Instructed Peptide Assembly Favored by Preorganization for Cancer Cell Membrane Engineering.

Innovative methods for engineering cancer cell membranes promise to manipulate cell-cell interactions and boost cell-based cancer therapeutics. Here, we illustrate an in situ approach to selectively modify cancer cell membranes by employing an enzyme-instructed peptide self-assembly (EISA) strategy. Using three phosphopeptides (pY1, pY2, and pY3) targeting the membrane-bound epidermal growth factor receptor (EGFR) and differing in just one phosphorylated tyrosine, we reveal that site-specific phosphorylation patterns in pY1, pY2, and pY3 can distinctly command their preorganization levels, self-assembling kinetics, and spatial distributions of the resultant peptide assemblies in cellulo. Overall, pY1 is the most capable of producing preorganized assemblies and shows the fastest dephosphorylation reaction in the presence of alkaline phosphatase (ALP), as well as the highest binding affinity for EGFR after dephosphorylation. Consequently, pY1 exhibits the greatest capacity to construct stable peptide assemblies on cancer cell membranes with the assistance of both ALP and EGFR. We further use peptide-protein and peptide-peptide co-assembly strategies to apply two types of antigens, namely ovalbumin (OVA) protein and dinitrophenyl (DNP) hapten respectively, on cancer cell membranes. This study demonstrates a very useful technique for the in situ construction of membrane-bound peptide assemblies around cancer cells and implies a versatile strategy to artificially enrich cancer cell membrane components for potential cancer immunotherapy.

Swimming exercise ameliorates insulin resistance and nonalcoholic fatty liver by negatively regulating PPARγ transcriptional network in mice fed high fat diet.

BACKGROUND: Recent findings elucidated hepatic PPARγ functions as a steatogenic-inducer gene that activates de novo lipogenesis, and is involved in regulation of glucose homeostasis, lipid accumulation, and inflammation response. This study delved into a comprehensive analysis of how PPARγ signaling affects the exercise-induced improvement of insulin resistance (IR) and non-alcoholic fatty liver disease (NAFLD), along with its underlying mechanism. METHODS: Chronic and acute swimming exercise intervention were conducted in each group mice. IR status was assessed by GTT and ITT assays. Serum inflammatory cytokines were detected by Elisa assays. PPARγ and its target genes expression were detected by qPCR assay. Relative protein levels were quantified via Western blotting. ChIP-qPCR assays were used to detect the enrichment of PPARγ on its target genes promoter. RESULTS: Through an exploration of a high-fat diet (HFD)-induced IR and NAFLD model, both chronic and acute swimming exercise training led to significant reductions in body weight and visceral fat mass, as well as hepatic lipid accumulation. The exercise interventions also demonstrated a significant amelioration in IR and the inflammatory response. Meanwhile, swimming exercise significantly inhibited PPARγ and its target genes expression induced by HFD, containing CD36, SCD1 and PLIN2. Furthermore, swimming exercise presented significant modulation on regulatory factors of PPARγ expression and transcriptional activity. CONCLUSION: The findings suggest that swimming exercise can improve lipid metabolism in IR and NAFLD, possibly through PPARγ signaling in the liver of mice.

Visible Light Generation of a Microsecond Long-Lived Potent Reducing Agent.

Photoexcitation of molecular radicals can produce strong reducing agents; however, the limited lifetimes of the doublet excited states preclude many applications. Herein, we propose and demonstrate a general strategy to translate a highly energetic electron from a doublet excited state to a ZrO2 insulator, thereby increasing the lifetime by about 6 orders of magnitude while maintaining a reducing potential less than -2.4 V vs SCE. Specifically, red light excitation of a salicylic acid modified perylene diimide radical anion PDI•- anchored to a ZrO2 insulator yields a ZrO2(e-)|PDI charge separated state with an ∼10 µs lifetime in 23% yield. The ZrO2(e-)s were shown to drive CO2 → CO reduction with a Re catalyst present in micromolar concentrations. More broadly, this strategy provides new opportunities to reduce important reagents and catalysts at low concentrations through diffusional electron transfer.

Machine Learning-Based Model for Predicting Prolonged Mechanical Ventilation in Patients with Congestive Heart Failure.

BACKGROUND: Mechanical ventilation (MV) is widely used to relieve respiratory failure in patients with congestive heart failure (CHF). Prolonged MV (PMV) is associated with a poor prognosis. We aimed to establish a prediction model based on machine learning (ML) algorithms for the early identification of patients with CHF requiring PMV. METHODS: Twelve commonly used ML algorithms were used to build the prediction model. The least absolute shrinkage and selection operator (LASSO) regression was employed to select the key features. We examined the area under the curve (AUC) statistics to evaluate the prediction performance. Data from another database were used to conduct external validation. RESULTS: We screened out 10 key features from the initial 65 variables via LASSO regression to improve the practicability of the model. The CatBoost model showed the best performance for predicting PMV among the 12 commonly used ML algorithms, with favorable discrimination (AUC = 0.790) and calibration (Brier score = 0.154). Moreover, hospital mortality could be accurately predicted using the CatBoost model as well (AUC = 0.844). In the external validation, the CatBoost model also showed satisfactory prediction performance (AUC = 0.780), suggesting certain generalizability of the model. Finally, a nomogram with risk classification of PMV was shown in this study. CONCLUSION: The present study developed and validated a CatBoost model, which could accurately predict PMV in mechanically ventilated patients with CHF. Moreover, this model has a favorable performance in predicting hospital mortality in these patients.

A novel lncRNA FPASL regulates fibroblast proliferation via the PI3K/AKT and MAPK signaling pathways in hypertrophic scar.

Hypertrophic scar is a problem for numerous patients, especially after burns, and is characterized by increased fibroblast proliferation and collagen deposition. Increasing evidence demonstrates that lncRNAs contribute to the development and progression of various diseases. However, the function of lncRNAs in hypertrophic scar formation remains poorly characterized. In this study, a novel fibroblast proliferation-associated lncRNA, named lncRNA FPASL (MSTRG.389905.1), which is mainly localized in the cytoplasm, is found to be downregulated in hypertrophic scar, as detected by lncRNA microarray and qRT-PCR. The full-length FPASL is characterized and further investigation confirms that it has no protein-coding potential. FPASL knockdown in fibroblasts triggers fibroblast proliferation, whereas overexpression of FPASL directly attenuates the proliferation of fibroblasts. Furthermore, target genes of the differentially expressed lncRNAs in hypertrophic scars and the matched adjacent normal tissues are enriched in fibroblast proliferation signaling pathways, including the PI3K/AKT and MAPK signaling pathways, as determined by GO annotation and KEGG enrichment analysis. We also demonstrate that knockdown of FPASL activates the PI3K/AKT and MAPK signaling pathways, and specific inhibitors of the PI3K/AKT and MAPK signaling pathways can reverse the proliferation of fibroblasts promoted by FPASL knockdown. Our findings contribute to a better understanding of the role of lncRNAs in hypertrophic scar and suggest that FPASL may act as a potential novel therapeutic target for hypertrophic scar.

Lattice-Confined Single-Atom Fe1 Sx on Mesoporous TiO2 for Boosting Ambient Electrocatalytic N2 Reduction Reaction.

Mimicking natural nitrogenase to create highly efficient single-atom catalysts (SACs) for ambient N2 fixation is highly desired, but still challenging. Herein, S-coordinated Fe SACs on mesoporous TiO2 have been constructed by a lattice-confined strategy. The extended X-ray absorption fine structure and X-ray photoelectron spectroscopy spectra demonstrate that Fe atoms are anchored in TiO2 lattice via the FeS2 O2 coordination configuration. Theoretical calculations reveal that FeS2 O2 sites are the active centers for electrocatalytic nitrogen reduction reaction (NRR). Moreover, the finite element analysis shows that confinement of opened and ordered mesopores can facilitate the mass transport and offer an enlarged active surface area for NRR. As a result, this catalyst delivers a favorable NH3 yield rate of 18.3 µg h-1 mgcat. -1 with a high Faradaic efficiency of 17.3 % at -0.20 V versus a reversible hydrogen electrode. Most importantly, this lattice-confined strategy is universal and can also be applied to Ni1 Sx @TiO2 , Co1 Sx @TiO2 , Mo1 Sx @TiO2 , and Cu1 Sx @TiO2 SACs. Our study provides new hints for the design and biomimetic synthesis of highly efficient NRR electrocatalysts.

TAM and MUSIC Approach for Impact-Source Localization under Deformation Conditions.

As an impact-source-localization technique, Lamb waves are commonly used to detect low-velocity impact in composite structures. However, the performance of Lamb waves is susceptible under deformation conditions. In this paper, a novel approach combined the Toeplitz approximation method (TAM) and multiple-signal classification (MUSIC) (TAM-MUSIC) to improve impact-source-localization (angle and distance in polar coordinates) accuracy under deformation conditions. The method divided a two-dimensional search of direction and distance into two one-dimensional searches. The impact direction was calculated by the TAM, which was introduced into the steering vector of MUSIC to estimate the distance by scanning the whole monitoring area. An epoxy laminate plate experiment showed that the phase and amplitude of uniform linear-array signals had different average plate curvature that led to poor impact-source-localization accuracy using the MUSIC method. TAM provided high-resolution direction-finding capability, suitable for the pretreatment of Lamb waves. Results showed that the present method, with a small amount of computation and low memory requirement, had higher location-estimation accuracy than that of traditional methods under deformation conditions.

Theoretical investigation of photonic generation of frequency quadrupling linearly chirped waveform with large tunable range.

To generate linearly chirped microwave signals with a large frequency tunable range, a photonic approach is proposed. Firstly, A dual-output dual-parallel Mach-Zehnder modulator (DPMZM) followed by the polarization beam combiner and an optical filter is utilized to generate orthogonally polarized ± second-order optical sidebands. Then a polarization modulator is employed to achieve the phase modulation of the two wavelengths. Finally, the balanced detection is applied to suppress the distortion and background noise. The key advantages of the proposed scheme are the central frequency multiplying operation and large frequency tunable range. Simulation results show that a linearly chirped pulse product with time-bandwidth as well as a compression ratio for the pulse of 11 and 9.3 respectively, and a peak-to-sidelobe ratio (PSR) of 7.4 dB is generated. The system has both good reconfigurability and tunability, its frequency can be continuously adjusted from about 10 GHz to as much as 50 GHz in principle.

Insecticidal Effects of Transgenic Maize Bt-Cry1Ab, Bt-Vip3Aa, and Bt-Cry1Ab+Vip3Aa against the Oriental Armyworm, Mythimna separata (Walker) in Southwest China.

The oriental armyworm, Mythimna separata (Walker), an important migratory pest of maize and wheat, is posing a severe threat to maize production in Asian countries. As source areas of spring-summer emigratory populations, the control of M. separata in southwestern China is of great significance for East Asian maize production. To assess the toxicity of Bt maize against the pest, bioassays of Bt-(Cry1Ab+Vip3Aa) maize (event DBN3601T), Bt-Cry1Ab maize (event DBN9936), and Bt-Vip3Aa maize (event DBN9501) were conducted in Yunnan province of southwest China. There were significant differences in insecticidal activity between the three Bt maize events, and DBN3601T presented the highest insecticidal role. The results also indicated that the insecticidal effect of various Bt maize tissues took an order in leaf > kernel > silk, which is highly consistent with the expression amounts of Bt insecticidal protein in leaf (69.69 ± 1.18 µg/g), kernel (11.69 ± 0.75 µg/g), and silk (7.32 ± 0.31 µg/g). In field trials, all larval population densities, plant damage rates, and leaf damage levels of DBN3601T maize were significantly lower than the conventional maize. This research indicated that the DBN3601T event had a high control efficiency against M. separata and could be deployed in southwest China for the management of M. separata.

mRNA-Seq of testis and liver tissues reveals a testis-specific gene and alternative splicing associated with hybrid male sterility in dzo.

Alternative splicing (AS) plays an important role in the co-transcription and post-transcriptional regulation of gene expression during mammalian spermatogenesis. The dzo is the male F1 offspring of an interspecific hybrid between a domestic bull (Bos taurus ♂) and a yak (Bos grunniens ♀) which exhibits male sterility. This study aimed to identify the testis-specific genes and AS associated with hybrid male sterility in dzo. The iDEP90 program and rMATS software were used to identify the differentially expressed genes (DEG) and differential alternative splicing genes (DSG) based on RNA-seq data from the liver (n = 9) and testis (n = 6) tissues of domestic cattle, yak, and dzo. Splicing factors (SF) were obtained from the AmiGO2 and the NCBI databases, and Pearson correlation analysis was performed on the differentially expressed SFs and DSGs. We focused on the testis-specific DEGs and DSGs between dzo and cattle and yak. Among the top 3,000 genes with the most significant variations between these 15 samples, a large number of genes showed testis-specific expression involved with spermatogenesis. Cluster analysis showed that the expression levels of these testis-specific genes were dysregulated during mitosis with a burst downregulation during the pachynema spermatocyte stage. The occurrence of AS events in the testis was about 2.5 fold greater than in the liver, with exon skipping being the major AS event (81.89% to 82.73%). A total of 74 DSGs were specifically expressed in the testis and were significantly enriched during meiosis I, synapsis, and in the piRNA biosynthesis pathways. Notably, STAG3 and DDX4 were of the exon skipping type, and DMC1 was a mutually exclusive exon. A total of 36 SFs were significantly different in dzo testis, compared with cattle and yak. DDX4, SUGP1, and EFTUD2 were potential SFs leading to abnormal AS of testis-specific genes in dzo. These results show that AS of testis-specific genes can affect synapsis and the piRNA biosynthetic processes in dzo, which may be important factors associated with hybrid male sterility in dzo.

The interspecific hybrid offspring of a domestic bull (Bos taurus) and a yak (Bos grunniens) display heterosis in meat and milk production. The hybrid offspring are particularly adaptable to the harsh environments of the Qinghai-Tibet Plateau. However, the male F1 to F3 offspring of this interspecies hybrid are infertile, and spermatogenesis is arrested at meiosis preventing the prolonged utilization of the benefits of heterosis. This study aimed to identify the testis-specific genes and alternative splicing (AS) associated with hybrid male sterility using RNA-Seq data from the liver and testis tissues of domestic cattle, yaks, and their F1 offspring (dzo). The expression of the testis-specific genes became disordered during mitosis and meiosis in dzo. Their testis-specific genes with AS events were enriched during synapsis and in the piRNA biosynthetic processes. In addition, we identified the potential splicing factors associated with abnormal testis-specific AS gene expression in dzo. These results reveal the important role of AS in the meiotic arrest of dzo.

Commonly used software tools produce conflicting and overly-optimistic AUPRC values.

The precision-recall curve (PRC) and the area under it (AUPRC) are useful for quantifying classification performance. They are commonly used in situations with imbalanced classes, such as cancer diagnosis and cell type annotation. We evaluated 10 popular tools for plotting PRC and computing AUPRC, which were collectively used in >3,000 published studies. We found the AUPRC values computed by the tools rank classifiers differently and some tools produce overly-optimistic results.

Commonly used software tools produce conflicting and overly-optimistic AUPRC values.

The precision-recall curve (PRC) and the area under the precision-recall curve (AUPRC) are useful for quantifying classification performance. They are commonly used in situations with imbalanced classes, such as cancer diagnosis and cell type annotation. We evaluate 10 popular tools for plotting PRC and computing AUPRC, which were collectively used in more than 3000 published studies. We find the AUPRC values computed by the tools rank classifiers differently and some tools produce overly-optimistic results.