Reproducibility Of Experiments Research Articles

A versatile Hong–Ou–Mandel interference experiment in optical fiber for the undergraduate laboratory

Hong–Ou–Mandel (HOM) interference is a quantum optics laboratory experiment that has recently become more accessible to undergraduate students. The experiment consists of two identical photons simultaneously entering a nonpolarizing beamsplitter, where the photon wavefunctions interfere. As a result, the photon pair exits from the same beamsplitter output, whereas classically, the two photons are equally likely to exit from the two different outputs. This effect is demonstrated by observation of a dip in coincidence counts measured between the two beamsplitter outputs. Due to the precision needed to achieve photon indistinguishability, the setup and alignment of this experiment is often considered to be too difficult and time consuming to be appropriate for the undergraduate instructional laboratory. Here, we present an alternative optical-fiber-based apparatus that gives a consistently reproducible experiment. In our approach, quantum interference occurs within a fused-fiber coupler, instead of within a traditional beamsplitter. We use a commercially available fiber-coupled biphoton source that requires minimal alignment and increases the interference coherence length. In addition, our biphoton source provides direct temperature-based control of the frequency degeneracy of the photon pairs, allowing students to investigate physical properties of HOM interference such as coherence length and interference visibility. Through use of standard optomechanical parts, combined with the commercially available fiber-integrated biphoton source, our apparatus is positioned midway between a completely built-from-scratch and a pre-aligned setup, making it ideal for the advanced instructional laboratory.

A novel method for assessing the shedding of fibre in forensic science: Investigating the effects of washing.

The evaluation of the shedding capacity of a garment is crucial in forensic analysis to understand fibre transfer mechanisms during contact activities. While adhesive tapes are commonly used, the lack of standardised pressure application -often done manually- poses a challenge. In addition, while previous studies have examined the effects of washing on fibre evidence, there is a notable absence in the literature regarding its impact on garment shedding capacity. This study aims to address these gaps by proposing a practical method to assess garment shedding capacity. Conventional tape lifting experiments, involving manual pressure application, were conducted for comparison with the novel method proposed in this study. Controlled conditions for reproducible experiments were achieved using a cost-effective friction tester and automated data collection through photography and ImageJ image processing software. Through controlled simulations, this study seeks to examine the relationship between garment shedding capacity, fibre transfer dynamics, and the effects of textile washing during laundry cycles.

Effectively simplified Adriamycin-induced chronic kidney disease mouse model: Retro-orbital vein injection versus tail-vein injection.

This study aimed to investigate the impact of administration routes in establishing the Adriamycin (ADR)-induced chronic kidney disease (CKD) model. Using BALB/c mice, we compared the effects of conventional tail-vein injection (TV10, 10 mg/kg) to those of retro-orbital sinus (orbital vein) injection (OV10, 10 mg/kg; OV8, 8 mg/kg). The results indicated that the OV10 group exhibited CKD pathology similar to the TV10 group, with both groups demonstrating significantly higher urinary albumin/creatinine ratio (p < 0.05), tubular injury (p < 0.05), and degree of renal fibrosis (p < 0.05) than the OV8 group. No significant differences were observed between the OV10 and TV10 groups in urinary albumin/creatinine ratio, tubular injury, and degree of renal fibrosis. These findings demonstrated that retro-orbital administration of 10 mg/kg ADR induces comparable effects to conventional tail-vein administration. This technique's technical simplicity may improve experimental efficiency, reproducibility, and animal welfare in CKD research. In conclusion, this study validates the utility of retro-orbital injection in CKD model establishment, demonstrating its potential to standardize and improve the reliability of future CKD research protocols.

Pervasive Games for Sexual Health Promotion: Scoping Literature Review.

Serious games play a fundamental role in promoting safe sexual behaviors. This medium has great potential for promoting healthy behaviors that prevent potential risk factors, such as sexually transmitted infections, and promote adherence to sexual health treatments, such as antiretroviral therapy. The ubiquity of mobile devices enhances access to such tools, increasing the effectiveness of video games as agents of change. In this scoping review, we aimed to (1) identify the extent to which pervasive games have been used in the field of sexual health, (2) determine the theories used in the design and evaluation of pervasive games for sexual health, (3) identify the methods used to evaluate pervasive games for sexual health, and (4) explore the reported benefits of using pervasive games for sexual health. Following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) methodology, we conducted a comprehensive literature search in the Web of Science, Scopus, IEEE Xplore, and ACM databases for articles published between January 1, 2000, and August 4, 2024. Included articles were published in English between 2000 and 2024 and involved the design, implementation, or evaluation of a ubiquitous video game focused on promoting safe sexual behaviors, with qualitative and/or quantitative results based on theory-based techniques and ubiquitous technologies. Review articles, conference papers, or books without available data or quantitative or qualitative results were excluded. We screened 521 of 612 articles (85.1%) after removing duplicates. After the title and abstract review, 51 (9.8%) articles were assessed for eligibility, and 30 (5.8%) articles meeting the criteria were studied and evaluated in depth. The results suggested that the use of pervasive video games has a positive impact on promoting safe sexual behaviors. This is enhanced by the effectiveness of theory-based techniques and the use of mobile technologies as developmental factors that drive the gaming experience. The results indicated that this domain is a growing field that should not be ignored. The literature showed that pervasive video games have been effective in promoting safe sexual behaviors. Substantial growth has been seen in scientific community interest in researching this domain; nevertheless, there is still much to work on. In this context, we advocate for the standardization of design, implementation, and experimentation as essential phases in creating video game experiences. These 3 fundamental aspects are critical in the development of video game-based studies to ensure the reproducibility of experiments.

Microsensor systems for cell metabolism - from 2D culture to organ-on-chip (2019-2024).

Cell cultures, organs-on-chip and microphysiological systems become increasingly relevant as in vitro models, e.g., in drug development, disease modelling, toxicology or cancer research. It has been underlined repeatedly that culture conditions and metabolic cues have a strong or even essential influence on the reproducibility and validity of such experiments but are often not appropriately measured or controlled. Here we review microsensor systems for cell metabolism for the continuous measurement of culture conditions in microfluidic and lab-on-chip platforms. We identify building blocks, features and essential advantages to underline the relevance of these systems and to derive appropriate requirements for development and practical use. We discuss different formats and geometries of cell culture, microfluidics and the resulting consequences for sensor placement, as the prerequisite for understanding the various approaches and classification of the systems. The major chemical and biosensors based on electrochemical and optical principles are discussed for general understanding and to contextualize current developments. We then review selected recent sensor systems with real-world implementations of sensing in cell cultures and organs-on-chip, employing a helpful characterization. That includes formats and cell models, microfluidic systems and sensor types applied in static and dynamic monitoring of 2D and 3D cell cultures, as well as single spheroids. We discuss notable advances, particularly with respect to sensor performance and the demonstration of long-term continuous measurements. We outline current approaches to system fabrication technologies, material choice, and interfacing, and comment on recent trends. Finally, we conclude with critical remarks on the current state of sensors in cell culture monitoring and identify avenues for future improvements for both developers and users of such systems, which will lead to better and more predictive in vitro models.

Functionalization of Polyethylene Terephthalate (PETE) Membranes for the Enhancement of Cellular Adhesion in Organ-on-a-Chip Devices.

Experimental reproducibility in organ-on-chip (OOC) devices is a challenging issue, mainly caused by cell adhesion problems, as OOC devices are made of bioinert materials not suitable for natural cellularization of their surfaces. To improve cell adhesion, several surface functionalization techniques have been proposed, among which the simple use of an intermediate layer of adsorbed proteins has become the preferred one by OOC users. This way, the cells use surface receptors to adhere to the adsorbed proteins, which are in turn attached to the surface. However, as protein adsorption is based on weak electrostatic bonding between the coating proteins and the substrate, this method produces suboptimal results: as the weak electrostatic bonds break, cells detach, leading to poor, heterogeneous cellularization. To solve this problem, we present a surface functionalization method for polyethylene terephthalate (PETE) membranes, commonly used in multilayer organ-on-chip devices to support cellular layers. This protocol involves hydrolyzation of the membrane, followed by (3-dimethylaminopropyl) carbodiimide (EDC) and N-hydroxysuccinimide (NHS) activation, resulting in covalent bonding between the membrane and coating proteins, much stronger than the weak electrostatic bonding provided by simple adsorption. As evaluation, we first measured the effect of the functionalization protocol in the morphological and mechanical integrity of the membranes. Next, we confirmed protein coating efficiency using the ζ potential and surface tension of the functionalized membranes coated with collagen type I, polylysine, gelatin, albumin, fetal bovine serum (FBS), and Matrigel. Finally, we showed that our method significantly improves the attachment of epithelial (A549) and endothelial (EA.hy926) cell lines under static conditions, especially in collagen-coated membranes, which were further tested under dynamic conditions, showing statistically significant improvement in cell attachment compared to uncoated or collagen-adsorbed only membranes.

DECODE enables high-throughput mapping of antibody epitopes at single amino acid resolution.

Antibodies are extensively used in biomedical research, clinical fields, and disease treatment. However, to enhance the reproducibility and reliability of antibody-based experiments, it is crucial to have a detailed understanding of the antibody's target specificity and epitope. In this study, we developed a high-throughput and precise epitope analysis method, DECODE (Decoding Epitope Composition by Optimized-mRNA-display, Data analysis, and Expression sequencing). This method allowed identifying patterns of epitopes recognized by monoclonal or polyclonal antibodies at single amino acid resolution and predicted cross-reactivity against the entire protein database. By applying the obtained epitope information, it has become possible to develop a new 3D immunostaining method that increases the penetration of antibodies deep into tissues. Furthermore, to demonstrate the applicability of DECODE to more complex blood antibodies, we performed epitope analysis using serum antibodies from mice with experimental autoimmune encephalomyelitis (EAE). As a result, we were able to successfully identify an epitope that matched the sequence of the peptide inducing the disease model without relying on existing antigen information. These results demonstrate that DECODE can provide high-quality epitope information, improve the reproducibility of antibody-dependent experiments, diagnostics and therapeutics, and contribute to discover pathogenic epitopes from antibodies in the blood.

Optimized separation of astaxanthin stereoisomers from microbial sources using chiral HPLC.

Astaxanthin (AST) is a high-value antioxidant, and its efficient isolation and utilization are challenging owing to the presence of different stereoisomers from various sources. In the present study, a semi-preparative HPLC method for the efficient separation of AST stereoisomers using a Chiralpak IC chiral column with good loading capacity and chiral recognition ability was successfully developed. The mobile phase was methanol-methyl tert-butyl ether (90 : 10, v/v), with a flow rate of 3.06 mL min-1 and a maximum injection volume of 0.32 mg. The results indicated that the purity of all-trans AST was 97.9% for Haematococcus pluvialis and 97.5% for Phaffia rhodozyma. Additionally, molecular weights and fragmentation patterns analyzed using mass spectrometry were consistent with those of all-trans AST. Linearity validation and reproducibility experiments revealed that all calibration curves had coefficients of determination (R2) greater than 0.999 and a relative standard deviation (RSD) of <3.8%. This is because all-trans AST stereoisomers could undergo specific rotations or spins due to π-π interactions, hydrogen bonding, and inclusion interactions. This process allowed the successful separation of the three all-trans AST optical isomers and provides a theoretical basis for large-scale preparation of all-trans AST stereoisomers from different sources.

Standardizing experimental approaches to investigate interactions between bacteria and ectomycorrhizal fungi.

Bacteria and ectomycorrhizal fungi (EcMF) represent two of the most dominant plant root-associated microbial groups on Earth, and their interactions continue to gain recognition as significant factors that shape forest health and resilience. Yet we currently lack a focused review that explains the state of bacteria-EcMF interaction research in the context of experimental approaches and technological advancements. To these ends, we illustrate the utility of studying bacteria-EcMF interactions, detail outstanding questions, outline research priorities in the field, and provide a suite of approaches that can be used to promote experimental reproducibility, field advancement, and collaboration. Though this review centers on the ecology of bacteria, EcMF, and trees, it by default offers experimental and conceptual insights that can be adapted to various subfields of microbiology and microbial ecology.

Cellular Attachment Assays to Dissect Molecular Drivers of Epidermal Cell Function.

Cell attachment is the process by which cells interact with their environment, including neighboring cells and the extracellular matrix (ECM). Attachment plays a critical role in maintaining skin integrity, promoting wound healing, and facilitating cellular communication in epidermal cells, such as keratinocytes. However, the many different factors that can influence this mechanism make it challenging to recapitulate in cellular models. The overlap between attachment and adhesion mechanisms both physiologically and methodologically further complicate the production of cellular models. Here, we present a flexible, quantitative, and cost-effective tool for studying epidermal attachment under various conditions. We provide optimized starting conditions for several different adaptations of the core protocol and provide approaches for quantitative, reproducible data that can be performed in most laboratories. This assay enhances experimental reproducibility and enables a targeted approach to studying epidermal biology. This approach offers researchers an improved tool for dissecting the molecular events in cell attachment and advancing skin biology research.

Performance and Reproducibility of Large Language Models in Named Entity Recognition: Considerations for the Use in Controlled Environments.

Recent artificial intelligence (AI) advances can generate human-like responses to a wide range of queries, making them a useful tool for healthcare applications. Therefore, the potential use of large language models (LLMs) in controlled environments regarding efficacy, reproducibility, and operability will be of paramount interest. We investigated if and how GPT 3.5 and GPT 4 models can be directly used as a part of a GxP validated system and compared the performance of externally hosted GPT 3.5 and GPT 4 against LLMs, which can be hosted internally. We explored zero-shot LLM performance for named entity recognition (NER) and relation extraction tasks, investigated which LLM has the best zero-shot performance to be used potentially for generating training data proposals, evaluated the LLM performance of seven entities for medical NER in zero-shot experiments, selected one model for further performance improvement (few-shot and fine-tuning: Zephyr-7b-beta), and investigated how smaller open-source LLMs perform in contrast to GPT models and to a small fine-tuned T5 Base. We performed reproducibility experiments to evaluate if LLMs can be used in controlled environments and utilized guided generation to use the same prompt across multiple models. Few-shot learning and quantized low rank adapter (QLoRA) fine-tuning were applied to further improve LLM performance. We demonstrated that zero-shot GPT 4 performance is comparable with a fine-tuned T5, and Zephyr performed better than zero-shot GPT 3.5, but the recognition of product combinations such as product event combination was significantly better by using a fine-tuned T5. Although Open AI launched recently GPT versions to improve the generation of consistent output, both GPT variants failed to demonstrate reproducible results. The lack of reproducibility together with limitations of external hosted systems to keep validated systems in a state of control may affect the use of closed and proprietary models in regulated environments. However, due to the good NER performance, we recommend using GPT for creating annotation proposals for training data as a basis for fine-tuning.

Timing of standard chow exposure determines the variability of mouse phenotypic outcomes and gut microbiota profile.

Standard chow diets influence reproducibility in animal model experiments because chows have different nutrient compositions, which can independently influence phenotypes. However, there is little evidence of the role of timing in the extent of variability caused by chow exposure. Here we measured the impact of different diets (5V5M, 5V0G, 2920X and 5058) and timing of exposure (adult exposure (AE), lifetime exposure (LE) and developmental exposure (DE)) on growth and development, metabolic health indicators and gut bacterial microbiota profiles across genetically identical C57BL/6J mice. Diet drove differences in macro- and micronutrient intake for all exposure models. AE had no effect on phenotypic outcomes. However, LE mice exhibited significant sex-dependent diet effects on growth, body weight and body composition. LE effects were mostly absent in the DE model, where mice were exposed to chow differences only from conception to weaning. Both AE and LE models exhibited similar diet-driven beta diversity profiles for the gut bacterial microbiota, with 5058 diet driving the most distinct profile. However, compared with AE, LE effects on beta diversity were sex dependent, and LE mice exhibited nine times more differentially abundant bacterial genera, the majority of which were inversely affected by 2920X and 5058 diets. Our findings demonstrate that LE to different chow diets has the greatest impact on the reproducibility of several experimental measures commonly used in preclinical mouse model studies. Importantly, weaning mice from different diets onto the same diet for maturation may be an effective way to reduce unwanted phenotypic variability among experimental models.

Development of a Multiple Temperature Sensors Device for the Characterization, Control and Monitoring of Microbiological Incubators.

An increasing number of projects require the precise knowledge and control of thermal conditions within the study system and their temporal evolution. This is particularly critical for equipment such as laboratory ovens and microbiological incubators, which are essential in various fields of chemistry and microbiology areas. These devices allow and facilitate the execution of experimental work in controlled environments, leading to reproducible experiments. This work presents a methodology for assembling and calibrating a highly accessible and low-cost data logger equipped with multiple temperature sensors. The final calibrated dispositive is straightforward to construct and allows the simultaneous and independent temperature measurement from multiple positions within the same system, which is then applied to the study, characterization, control, and monitoring of the internal thermal behavior of a laboratory oven dedicated to microbiological agents' cultivation. This approach ensures, through a robust methodology, a precise characterization by quantitative methods that allows objective decision making in the management and control of the temperature inside the system. Additionally, the device is suitable for extension and application in diverse research environments by modifying the sensor calibration to achieve a desired temperature range or number of measurement units, representing a potential work tool for laboratory systems.

Textflows: an open science NLP evaluation approach

AbstractWeb-based platforms offer suitable experimental environments enabling the construction and reuse of natural language processing (NLP) pipelines. However, systematic evaluation of NLP tools in an open science web-based setting is still a challenge, as suitable experimental environments for the construction and reuse of NLP pipelines are still rare. This paper presents TextFlows, an open-source web-based platform, which enables user-friendly construction, sharing, execution, and reuse of NLP pipelines. It demonstrates that TextFlows can be easily used for systematic evaluation of new NLP components by integrating seven publicly available open-source part of speech (POS) taggers from popular NLP libraries, and evaluating them on six annotated corpora. The integration of new tools into TextFlows supports tools reuse, while the use of precomposed algorithm comparison and evaluation workflows supports experiment reproducibility and testing of future algorithms in the same experimental environment. Finally, to showcase the variety of evaluation possibilities offered in the TextFlows platform, the influence of various factors, such as the training corpus length and the use of pre-trained models, have been tested.

Robots and Dance: A Promising Young Alchemy

Research at the intersection of robots and dance promises to create vehicles for expression that enable new creative pursuits and allow robots to function better, especially in human-facing scenarios. Moving this research beyond fringe spectacle and establishing it as a serious, systematic field—a proper subdiscipline of both robotics and dance—will require answering a key question: How does dance advance the fundamentals of robotics, and vice versa? Focusing on the former, this article offers glimpses of this new field with examples of meaningful contributions to control, robotics, and autonomous systems, such as novel actuator designs, improved sensing systems, salient motion profiles for robots, reproducible experiment designs, and new theories of motion derived from the study of dance. It also poses two grand challenges for the emerging field of choreobotics: developing a robust symbolic system for representing bodily action and establishing rich, repeatable testing environments for human–robot interaction.

Evaluating Study Design Rigor in Preclinical Cardiovascular Research: A Replication Study.

Methodological rigor remains a priority in preclinical cardiovascular research to ensure experimental reproducibility and high-quality research. Limited reproducibility diminishes the translation of preclinical discoveries into medical practice. In addition, lack of reproducibility fosters uncertainty in the public's acceptance of reported research results. We evaluated the reporting of methodological practices in preclinical cardiovascular research studies published in leading scientific journals by screening articles for the inclusion of the following study design elements (SDEs): considering sex as a biological variable, randomization, blinding, and sample size power estimation. We screened for these SDEs across articles regarding preclinical cardiovascular research studies published between 2011 and 2021. We replicated and extended a study published in 2017 by Ramirez et al. We hypothesized a higher SDE inclusion across preclinical studies over time, that preclinical studies that include human and animal substudies within the same study will exhibit greater SDE inclusion than animal-only preclinical studies, and that a difference exists in SDE usage between large and small animal models. SDE inclusion was low; with 15.2% of animal-only studies including both sexes as a biological variable, 30.4% including randomization, 32.1% including blinding, and 8.2% including sample size estimation. The incorporation of SDEs did not significantly increase over the ten-year timeframe in the screened articles. Randomization and sample size estimation differed significantly between animal and human substudies (corrected p=1.85e-05 and corrected p=3.81e-07, respectively.). Evidence of methodological rigor varies depending on the study type and model organisms used. From 2011-2021, SDE reporting within preclinical studies has not increased, suggesting more work is needed to foster the inclusion of rigorous study design elements in cardiovascular research.

The Genus Commiphora: An Overview of Its Traditional Uses, Phytochemistry, Pharmacology, and Quality Control.

Myrrh is the resinous substance secreted by plants of the genus Commiphora. In traditional Chinese medicine, Ayurvedic medicine, and traditional Arabic medicine, myrrh is regarded as an important medicinal material, widely used in the treatment of trauma, arthritis, hyperlipidemia, and other diseases. This review explores the evolving scientific understanding of the genus Commiphora, covering facets of ethnopharmacology, phytochemistry, pharmacology, artificial cultivation, and quality control. In particular, the chemical constituents and pharmacological research are reviewed. More than 300 types of secondary metabolites have been identified through phytochemical studies of this genus. Guggulsterone is a bioactive steroid isolated mainly from Commiphora mukul. The two isomers, Z- and E-guggulsterone, have shown a wide range of in vitro and in vivo pharmacological effects, including anti-proliferation, antioxidant, anti-inflammatory, and antibacterial. However, the current scientific research on quality control of medicinal materials and identification of original plants is insufficient, which limits the reproducibility and accuracy of biological activity evaluation experiments. Therefore, the establishment of analytical protocols and standardization of extracts is an important step before biological evaluation. At the same time, in order to find more bioactive substances, it is necessary to strengthen the research on the stems, barks, and leaves of this genus. The sources used in this study include PubMed, CNKI, Web of Science, Google Scholar, and other databases, as well as multinational pharmacopoeias, ancient books of traditional medicine, herbal classics, and modern monographs.

Optimized primary dorsal root ganglion cell culture protocol for reliable K+ current patch-clamp recordings

DRG primary neuron cultures, derived from rodents, closely mimic properties of sensory neurons in vivo and are highly useful for studying pain and neurological disorders. These cultures are pivotal in patch-clamp electrophysiology for sensory neuron properties analysis. A detailed, replicable protocol in scientific research ensures experiment accuracy and reproducibility. This paper provides comprehensive details for replicating the protocol and achieving consistent results in primary DRG cell culture as used for patch-clamp recordings. We outlined a comprehensive protocol for establishing primary DRG cell culture, optimized for improved gigaseal formation in whole-cell patch-clamp recordings. Additionally, we conducted a simulation study focused on recording macroscopic K+ channels. The findings established an optimized novel protocol that works reliably for whole-cell patch-clamp recordings and data analysis using primary DRG cells prepared as described in this publication. The details for the protocol in the literature are dispersed across various publications, making it challenging to find a comprehensive summary in one source. This study confirms, for the first time, the efficacy of using fewer protocol steps, which reduces stress and variability in obtaining suitable cells for patch-clamp recordings compared to existing methods in the literature. Given the challenges posed by the dissociation process of primary DRG cells and the importance of comprehensive method documentation in the literature, the protocol presented provides improved and consistent applications of primary DRG cell culture in patch-clamp recordings.

Codon and Signal Peptide Optimization to Enhance Therapeutic Antibody Production from CHO Cells.

A simple and rapid method for reproducible transient expression experiments to quickly identify the best conditions before the development of a stable CHO cell line is described, using adalimumab as a model antibody. Firstly, the signal peptide (SP) is optimized using different in silico programs, of which those having a D score of more than 0.85 will be selected. A list of useful SPs is provided. Next, guidelines for selecting the codon optimization algorithm is provided based on aCAI value of more than 0.95. After that, the codon optimized genes of the heavy chain (HC) and light chain (LC) of an antibody with different SPs are engineered into an expression vector and the two recombinant plasmids are co-transfected into a CHO cell line of interest. The antibody titer, specific productivity (Qp), and viable cell density (VCD) are then checked by ELISA and flow cytometry analysis. The best SP pairs can then be used for the development of stable CHO cell lines.

Improved protein interaction models predict differences in complexes between human cell lines.

The interactions of proteins to form complexes play a crucial role in cell function. Data on protein-protein or pairwise interactions (PPI) typically come from a combination of sample separation and mass spectrometry. Since 2010, several extensive, high-throughput mass spectrometry-based experimental studies have dramatically expanded public repositories for PPI data and, by extension, our knowledge of protein complexes. Unfortunately, challenges of limited overlap between experiments, modality-oriented biases, and prohibitive costs of experimental reproducibility continue to limit coverage of the human protein assembly map, both underscoring the need for and spurring the development of relevant computational approaches. Here, we present a new method for predicting the strength of protein interactions. It addresses two important issues that have limited past PPI prediction approaches: incomplete feature sets and incomplete proteome coverage. For a given collection of protein pairs, we fused data from heterogeneous sources into a feature matrix and identified the minimal set of feature partitions for which a non-empty set of protein pairs had complete values. For each such feature partition, we trained a classifier to predict PPI probabilities. We then calculated an overall prediction for a given protein pair by weighting the probabilities from all models that applied to that pair. Our approach accurately identified known and highly probable PPI, far exceeding the performance of current approaches and providing more complete proteome coverage. We then used the predicted probabilities to assemble complexes using previously-described graph-based tools and clustering algorithms and again obtained improved results. Lastly, we used features for three human cell lines to predict PPI and complex scores and identified complexes predicted to differ between those cell lines.