Instance maps as an organising concept for complex experimental workflows as demonstrated for (nano)material safety research

• Benjamin Punz,
• Maja Brajnik,
• Joh Dokler,
• Jaleesia D. Amos,
• Litty Johnson,
• Katie Reilly,
• Anastasios G. Papadiamantis,
• Amaia Green Etxabe,
• Lee Walker,
• Diego S. T. Martinez,
• Steffi Friedrichs,
• Klaus M. Weltring,
• Nazende Günday-Türeli,
• Claus Svendsen,
• Christine Ogilvie Hendren,
• Mark R. Wiesner,
• Martin Himly,
• Iseult Lynch and
• Thomas E. Exner

Beilstein J. Nanotechnol. 2025, 16, 57–77, doi:10.3762/bjnano.16.7

• for specific purposes and can modulate toxicity endpoints. Ideally, a nomenclature would include details on chemical identities of the nanoparticle’s core and surface, its transformation, where it is the transformed form of the nanomaterial that is evaluated, any impurities, and physical descriptors

The round-robin approach applied to nanoinformatics: consensus prediction of nanomaterials zeta potential

• Dimitra-Danai Varsou,
• Arkaprava Banerjee,
• Joyita Roy,
• Kunal Roy,
• Giannis Savvas,
• Haralambos Sarimveis,
• Ewelina Wyrzykowska,
• Mateusz Balicki,
• Tomasz Puzyn,
• Georgia Melagraki,
• Iseult Lynch and
• Antreas Afantitis

Beilstein J. Nanotechnol. 2024, 15, 1536–1553, doi:10.3762/bjnano.15.121

• ]. Another example of an IATA is the combination of predictions from two or more individual models under a consensus framework. Consensus models combine outputs from several individual models built upon different sets of descriptors and/or machine learning (ML) algorithms, leading to more trustworthy results

• generally more toxic than negatively charged particles of similar composition [28][29][30]. In fact, several in silico models for the ZP have been developed based on different theoretical and experimental descriptors employing a range of approaches, that is, quantitative structure–property/feature

• pristine and aged NPs, considering the size, coating, absolute electronegativity, and periodic table descriptors. Finally, advances of artificial intelligence (AI) have been also considered in the computational assessment of the ZP. Yan et al. [35] employed deep learning techniques and developed a

Integrating high-performance computing, machine learning, data management workflows, and infrastructures for multiscale simulations and nanomaterials technologies

• Fabio Le Piane,
• Mario Vozza,
• Matteo Baldoni and
• Francesco Mercuri

Beilstein J. Nanotechnol. 2024, 15, 1498–1521, doi:10.3762/bjnano.15.119

• of materials, thus facilitating the integration of diverse data sources and tools to develop predictive models under a structured assessment strategy. Among the broad range of tools available for supporting the development of digital twins of materials and the evaluation of molecular descriptors

• calculation of molecular descriptors that can be integrated into ML models [70]. Enalos NanoInformatics Cloud Platform is a web-based platform that allows users to design and build nanomaterials. It supports the calculation of molecular descriptors and the integration of these descriptors into ML models for

• suggests, silver, copper oxide, and titanium oxide [73]. ASCOT assists in the generation of high-quality digital twins of materials and the computation of relevant molecular descriptors. Nanotube Modeler is a software tool designed to create three-dimensional coordinates for various nanoscale carbon

AI-assisted models to predict chemotherapy drugs modified with C₆₀ fullerene derivatives

• Jonathan-Siu-Loong Robles-Hernández,
• Dora Iliana Medina,
• Katerin Aguirre-Hurtado,
• Marlene Bosquez,
• Roberto Salcedo and
• Alan Miralrio

Beilstein J. Nanotechnol. 2024, 15, 1170–1188, doi:10.3762/bjnano.15.95

• –COOH) were investigated with the protein CXCR7 as the molecular docking target. The research involved over 30 drugs and employed Pearson’s hard–soft acid–base theory and common QSAR/QSPR descriptors to build predictive models for the docking scores. Energetic descriptors were computed using quantum

• regression and IBM Watson artificial intelligence (AI), achieved mean absolute percentage errors below 12%, driven by AI-identified key variables. The predictive models included mainly quantitative descriptors collected from datasets as well as computed ones. In addition, a water-soluble fullerene was used

• descriptors, x1, x2, x3,..., xn, theoretically determined or measured by experiments [30]. A relationship f(x1, x2, x3,..., xn) can be defined to predict the activity or property of molecules after the evaluation of their quantitative descriptors. However, the QSAR/QSPR paradigm does not explain how to select

Introducing third-generation periodic table descriptors for nano-qRASTR modeling of zebrafish toxicity of metal oxide nanoparticles

• Supratik Kar and
• Siyun Yang

Beilstein J. Nanotechnol. 2024, 15, 1142–1152, doi:10.3762/bjnano.15.93

• zebrafish toxicity for 24 MONPs. Previously established 23 first- and second-generation periodic table descriptors, along with five newly proposed third-generation descriptors derived from the periodic table, were employed. Subsequently, to enhance the quality and predictive capability of the nano-QSTR

• model, a nano-quantitative read across structure–toxicity relationship (nano-qRASTR) model was created. This model integrated read-across descriptors with modeled descriptors from the nano-QSTR approach. The nano-qRASTR model, featuring three attributes, outperformed the previously reported simple QSTR

• models. Finally, the developed nano-qRASTR model was applied to predict toxicity data for an external dataset comprising 35 MONPs, addressing gaps in zebrafish toxicity assessment. Keywords: metal nanoparticles; metal oxide nanoparticles; nano-qRASTR; periodic table descriptors; QSAR; zebrafish

Atomistic insights into the morphological dynamics of gold and platinum nanoparticles: MD simulations in vacuum and aqueous media

• Evangelos Voyiatzis,
• Eugenia Valsami-Jones and
• Antreas Afantitis

Beilstein J. Nanotechnol. 2024, 15, 995–1009, doi:10.3762/bjnano.15.81

• multifaceted crystal surface, and their shape remains almost constant regardless of temperature variations. The smaller NPs have a smoother and more spherical surface, and their shape varies greatly with temperature. By studying the variation of nano-descriptors commonly employed in QSAR models, a qualitative

• information regarding the toxicity and reactivity of these NPs by monitoring the behaviour of nano-descriptors commonly employed in quantitative structure–activity relationship (QSAR) models and by measuring the water–NP energetic interactions. The extracted information from our simulations complements

• signifies the occurrence of a phase transition in a cluster of atoms. Additional atomic parameters are the average potential energy, force, and coordination number per atom. These quantities have also been employed as descriptors in nano-QSAR models to successfully predict the toxicity of NPs [73][74][75

Identification of structural features of surface modifiers in engineered nanostructured metal oxides regarding cell uptake through ML-based classification

• Indrasis Dasgupta,
• Totan Das,
• Biplab Das and
• Shovanlal Gayen

Beilstein J. Nanotechnol. 2024, 15, 909–924, doi:10.3762/bjnano.15.75

• descriptors were collected, including molecular weight (MW), n-octanol/water partition coefficient (ALogP), number of aromatic rings (nAR), number of rings (nR), number of rotatable bonds (nBonds), number of hydrogen bond donors (nHBDs), and the number of hydrogen bond acceptors (nHBAs) [36]. Extended

• using the fivefold cross-validation procedure. Additionally, the model’s quality was evaluated by looking at the receiver operating characteristic (ROC) plot as well as specificity, sensitivity, and accuracy values [40][41][42]. Development of other machine learning models Calculation of descriptors and

• data pre-treatment The training set of 88 and the test set of 21 surface modifiers from Bayesian classification analysis were used for the development of other machine learning models. Different classes of 2D descriptors were calculated using PaDEL-Descriptor [43]. The data pre-treatment tool (Data Pre

A review on the structural characterization of nanomaterials for nano-QSAR models

• Salvador Moncho,
• Eva Serrano-Candelas,
• Jesús Vicente de Julián-Ortiz and
• Rafael Gozalbes

Beilstein J. Nanotechnol. 2024, 15, 854–866, doi:10.3762/bjnano.15.71

• features. In this review, we survey the literature for existing predictive models for NMs and discuss the variety of calculated and experimental features used to define and describe NMs. In the light of this research, we propose a classification of the descriptors including those that directly describe a

• component of the nanoform (core, surface, or structure) and also experimental features (related to the nanomaterial’s behavior, preparation, or test conditions) that indirectly reflect its structure. Keywords: descriptors; nanomaterials; nano-QSAR; QSAR; toxicity; Introduction Computational techniques of

• structure of NMs [5][6][7]. The first described nano-QSAR model is from 2009 [8], but the number of relevant nano-QSAR models is growing significantly because new nanoscale descriptors are found [6], and more information on NMs is progressively generated, opening new ways of improving nano-QSARs. This is an

On the additive artificial intelligence-based discovery of nanoparticle neurodegenerative disease drug delivery systems

• Shan He,
• Julen Segura Abarrategi,
• Harbil Bediaga,
• Sonia Arrasate and
• Humberto González-Díaz

Beilstein J. Nanotechnol. 2024, 15, 535–555, doi:10.3762/bjnano.15.47

• and NP cytotoxicity datasets. Nevertheless, most of the AI/ML methods reported to date only consider the structural/molecular descriptors of the NDDs or NPs as input. Therefore, these methods exclude completely non-structural parameters, specifically experimental conditions of the assays, in order to

• target function by applying the vectors of descriptors for all cases Dk to use as the input variable in the ML model. The target function is commonly achieved by a mathematical conversion of the original theoretical or observed feature of the scheme under analysis [66][67][68]. In this IFPTML model, it

• the ChEMBL database [60][71][72]. It included the calculation of the vectors Dnk and Ddk of structural descriptors for all NPs and NDDs. In addition, we constructed the vectors cnj and cdj in order to list each label and assay condition for all preclinical assays of NPs and NDDs. Subsequently, we

Exploring the relationships between physiochemical properties of nanoparticles and cell damage to combat cancer growth using simple periodic table-based descriptors

• Joyita Roy and
• Kunal Roy

Beilstein J. Nanotechnol. 2024, 15, 297–309, doi:10.3762/bjnano.15.27

• treatment of cancer cells. To achieve this, QSPR modeling was first performed with 18 metal oxide (MeOx) NMs to measure their materials properties using periodic table-based descriptors. The features obtained were later applied for zeta potential calculation (imputation for sparse data) for MeOx NMs that

• lack such information. To further clarify the influence of the zeta potential on cell damage, a QSPR model was developed with 132 MeOx NMs to understand the possible mechanisms of cell damage. The results showed that zeta potential, along with seven other descriptors, had the potential to influence

• species. The most important criterion to improve nanoscale toxicity models is the selection of the appropriate structural descriptors of NPs. Periodic table-based descriptors have been a promising tool in predicting toxicity profiles and risk assessment of MeOx NPs with high predictivity and

Prediction of cytotoxicity of heavy metals adsorbed on nano-TiO₂ with periodic table descriptors using machine learning approaches

• Joyita Roy,
• Souvik Pore and
• Kunal Roy

Beilstein J. Nanotechnol. 2023, 14, 939–950, doi:10.3762/bjnano.14.77

• cause of cytotoxicity. To demonstrate the predictive ability of the developed nano-QSAR models, simple periodic table descriptors requiring low computational resources were utilized. The nano-QSAR models generated good R2 values (0.99–0.89), Q2 values (0.64–0.77), and Q2F1 values (0.99–0.71). Thus, the

• present work manifests that ML in conjunction with periodic table descriptors can be used to explore the features and predict unknown compounds with similar properties. Keywords: heavy metals; HK-2 cell; ML algorithm; periodic table descriptors; QSAR; Introduction Nanoparticles (NPs) have gained much

• AdaBoost) with periodic table descriptors for predicting the cytotoxicity, in terms of cell viability, of eight heavy metals adsorbed on nano-TiO2. Also, the best algorithm showing the most contributing features responsible for the toxicity to HK-2 (human kidney 2) cell has been determined. To the best

Zinc oxide nanostructures for fluorescence and Raman signal enhancement: a review

• Ioana Marica,
• Fran Nekvapil,
• Maria Ștefan,
• Cosmin Farcău and
• Alexandra Falamaș

Beilstein J. Nanotechnol. 2022, 13, 472–490, doi:10.3762/bjnano.13.40

• relevant descriptors of any SERS substrate is the signal enhancement factor (EF), which describes the enhancement of the Raman signal of target molecules when adsorbed on the SERS substrate relative to the conventional Raman signal of the same number of molecules. The EF is generally calculated according

Identifying diverse metal oxide nanomaterials with lethal effects on embryonic zebrafish using machine learning

• Richard Liam Marchese Robinson,
• Haralambos Sarimveis,
• Philip Doganis,
• Xiaodong Jia,
• Marianna Kotzabasaki,
• Christiana Gousiadou,
• Stacey Lynn Harper and
• Terry Wilkins

Beilstein J. Nanotechnol. 2021, 12, 1297–1325, doi:10.3762/bjnano.12.97

• evaluated using nested cross-validation on this dataset. Models were initially developed for both lethality endpoints using multiple descriptors representing the composition of the core, shell and surface functional groups, as well as particle characteristics. However, interestingly, the 24 hours post

• explored, yet both failed to boost predictive performance. Interestingly, it was found that comparable results to those originally obtained using multiple descriptors could be obtained using a model based upon a single, simple descriptor: the Pauling electronegativity of the metal atom. Since it is widely

• recognised that a variety of intrinsic and extrinsic nanomaterial characteristics contribute to their toxicological effects, this is a surprising finding. This may partly reflect the need to investigate more sophisticated descriptors in future studies. Future studies are also required to examine how robust

Prediction of Co and Ru nanocluster morphology on 2D MoS₂ from interaction energies

• Cara-Lena Nies and
• Michael Nolan

Beilstein J. Nanotechnol. 2021, 12, 704–724, doi:10.3762/bjnano.12.56

• there are two useful descriptors for 2D-vs-3D growth [48]: (1) If the metal–substrate interaction is more favourable than the metal–metal interaction, then 2D growth is preferred; and (2) if the total binding energy is more favourable than the cohesive energy of the bulk metal, then 2D growth is

• preferred. Predictions made using these descriptors can be used when deciding which metal–substrate combination will be suitable for a particular application where the shape of the metal is vital. Methods All calculations for this study were carried out with density functional theory (DFT) using the Vienna

Gas sorption porosimetry for the evaluation of hard carbons as anodes for Li- and Na-ion batteries

• Yuko Matsukawa,
• Fabian Linsenmann,
• Maximilian A. Plass,
• George Hasegawa,
• Katsuro Hayashi and
• Tim-Patrick Fellinger

Beilstein J. Nanotechnol. 2020, 11, 1217–1229, doi:10.3762/bjnano.11.106

• descriptors to the obtained capacities remains a scientific challenge. Keywords: alkaline-ion secondary battery; gas sorption porosimetry; hard carbon; irreversible capacity; ultramicroporosity; Introduction Lithium-ion battery (LIB)-based energy storage devices have been gaining high interest in the recent

• be carried out with CO2 or H2O sorption instead of N2 or Kr sorption. Of course, from the results of two RF carbons a trend cannot be concluded. However, it stands out that the slopes of assumed trends for pore descriptors derived from CO2 sorption (Figure 3b and Figure 4a) match the ones for the HT

• understanding and predicting the sodium and lithium storage capacities of HC anodes, and thereby elucidate the full potential of those materials for LIB and SIB technology. Conclusion We carried out a comparative study of porosity descriptors obtained by different gas sorption porosimetry techniques (Kr, N2

Heating ability of magnetic nanoparticles with cubic and combined anisotropy

• Nikolai A. Usov,
• Mikhail S. Nesmeyanov,
• Elizaveta M. Gubanova and
• Natalia B. Epshtein

Beilstein J. Nanotechnol. 2019, 10, 305–314, doi:10.3762/bjnano.10.29

• centers and the geometrical center of mass of the cluster. Interestingly, in contrast to usual 3D clusters, the dimension Df of a fractal cluster is typically a noninteger number. The fractal clusters with various fractal descriptors Df and kf were generated in this paper using the well known Filippov’s

• et al. algorithm [34]. Most of the calculations were performed for fractal clusters with Df = 1.9 and kf = 1.7 because these clusters are observed most often in biological media [24][25]. Similar results were obtained also for clusters with other fractal descriptors. As an example, Figure 1 shows the

• geometrical structure of fractal cluster with fractal descriptors Df = 1.9 and kf = 1.7 consisting of Np = 100 single-domain nanoparticles. The random space orientation of the ith spherical nanoparticle with cubic magnetic anisotropy is determined by a set of orthogonal unit vectors (e1i, e2i, e3i), i = 1, 2

Fabrication and photoactivity of ionic liquid–TiO₂ structures for efficient visible-light-induced photocatalytic decomposition of organic pollutants in aqueous phase

• Anna Gołąbiewska,
• Marta Paszkiewicz-Gawron,
• Aleksandra Sadzińska,
• Wojciech Lisowski,
• Ewelina Grabowska,
• Adriana Zaleska-Medynska and
• Justyna Łuczak

Beilstein J. Nanotechnol. 2018, 9, 580–590, doi:10.3762/bjnano.9.54

• issue in the field of IL–TiO2 composites, that is, which structural descriptors of ILs are crucial for the preparation of visible-light-active photocatalysts with desired morphology and properties and how to predict the properties of the IL–TiO2 material on the basis of the structure and properties of

Evaluating the toxicity of TiO₂-based nanoparticles to Chinese hamster ovary cells and Escherichia coli: a complementary experimental and computational approach

• Alicja Mikolajczyk,
• Natalia Sizochenko,
• Ewa Mulkiewicz,
• Anna Malankowska,
• Michal Nischk,
• Przemyslaw Jurczak,
• Seishiro Hirano,
• Grzegorz Nowaczyk,
• Adriana Zaleska-Medynska,
• Jerzy Leszczynski,
• Agnieszka Gajewicz and
• Tomasz Puzyn

Beilstein J. Nanotechnol. 2017, 8, 2171–2180, doi:10.3762/bjnano.8.216

• approach is based on defining mathematical dependencies between the variance in molecular structures, encoded by so-called molecular descriptors, and the variance in a given physicochemical property or biological (e.g., cytotoxicity) property in a set of compounds (“endpoints”) [5][25][26][27][28][29][30

• quantitatively describe the features of nanoparticles structure were considered as descriptors (Table 2). It is interesting to point out, that there was no significant linear correlation between the considered cytotoxicity and the descriptors (the Pearson correlation coefficient was lower than 0.5). To address

• this problem and to uncover the nonlinear relationship underlying measured data the Gaussian process approach was therefore used. The power of the Gaussian process approach, which uses lazy learning, is that it has an inherent ability to select the meaningful descriptors relevant to the endpoint of

Modeling adsorption of brominated, chlorinated and mixed bromo/chloro-dibenzo-p-dioxins on C₆₀ fullerene using Nano-QSPR

• Piotr Urbaszek,
• Agnieszka Gajewicz,
• Celina Sikorska,
• Maciej Haranczyk and
• Tomasz Puzyn

Beilstein J. Nanotechnol. 2017, 8, 752–761, doi:10.3762/bjnano.8.78

• adsorption energy values (ΔEads) for 32 Br/Cl dibenzo-p-dioxin congeners adsorbed on a C60 fullerene surface and carefully selected structural descriptors, we developed a Nano-QSPR model, employing a hybrid genetic algorithm, partial least squares linear regression (GA-PLS), as the modeling method. The

• developed Nano-QSPR model utilizes only four descriptors for predicting the adsorption energy values for 1,669 PXDD@C60 materials as follows: where #H is the number of hydrogen atoms in the dioxin molecule, TE is the total energy of the molecule, and D_x and D_y are the dipole moments of dioxin molecule

• on the cross-validation results (the lowest value of the root mean square error of cross validation, RMSECV). Four latent vectors (LVs), as a set of 4, were selected by the GA descriptors: #H, TE, D_x and D_y together explained 100% of the variance (58.08% + 40.42% + 1.30% + 0.20%) of the X-block and

The longstanding challenge of the nanocrystallization of 1,3,5-trinitroperhydro-1,3,5-triazine (RDX)

• Florent Pessina and
• Denis Spitzer

Beilstein J. Nanotechnol. 2017, 8, 452–466, doi:10.3762/bjnano.8.49

• -RDX composite deflagrates at a constant velocity of 2529 m/s. Those and other unpublished results from our laboratory confirm the drastic reduction of the critical diameter with the decrease of particle size. 1,3,5-Trinitroperhydro-1,3,5-triazine is also found under the following descriptors: cyclo

Improved biocompatibility and efficient labeling of neural stem cells with poly(L-lysine)-coated maghemite nanoparticles

• Igor M. Pongrac,
• Marina Dobrivojević,
• Lada Brkić Ahmed,
• Michal Babič,
• Miroslav Šlouf,
• Daniel Horák and
• Srećko Gajović

Beilstein J. Nanotechnol. 2016, 7, 926–936, doi:10.3762/bjnano.7.84

• six morphological descriptors, namely area, perimeter, Convex-Hull perimeter, equivalent diameter, roughness and circularity (Figure 1C). To further characterize the particle size distribution, number-equivalent diameter (Dn), weight-average diameter (Dw) and polydispersity index (PDI) were calculated

• micrographs of (A) PLL-γ-Fe2O3 and (B) nanomag®-D-spio nanoparticles. Insets show the corresponding electron diffraction patterns. (C) The nanoparticle morphology was characterized by measuring morphological descriptors. Area (A) and perimeter (P) of the analyzed particles were determined by counting the

• pixels using an image analysis software. Convex-Hull perimeter (C) and equivalent area were derived auxiliary descriptors. Key morphological descriptors were equivalent diameter (ED), circularity (CC) and roughness (RG). ED determined a diameter of a circle with the same area as the measured particle. CC

Nanoinformatics for environmental health and biomedicine

• Rong Liu and
• Yoram Cohen

Beilstein J. Nanotechnol. 2015, 6, 2449–2451, doi:10.3762/bjnano.6.253

• components calculated from nanoparticle size and surface properties using Kriging estimations [14]. Another contribution reports on the development of models to predict the cytotoxicity of PAMAM dendrimers using molecular descriptors [15]. Nanomaterials that have potential to cause disease (e.g., TiO2

Predicting cytotoxicity of PAMAM dendrimers using molecular descriptors

• David E. Jones,
• Hamidreza Ghandehari and
• Julio C. Facelli

Beilstein J. Nanotechnol. 2015, 6, 1886–1896, doi:10.3762/bjnano.6.192

• journal articles. The results indicate that data mining and machine learning can be effectively used to predict the cytotoxicity of PAMAM dendrimers on Caco-2 cells. Keywords: data mining; machine learning; molecular descriptors; poly(amido amine) dendrimers (PAMAM); Introduction In silico approaches

• potentially be expanded to other nanomaterials in the future. Results and Discussion Five different analyses were performed to classify a dendrimer as toxic or nontoxic using different combinations of molecular descriptors and experimental conditions. The first analysis utilized all the molecular descriptors

• available in MarvinSketch (see Experimental section and Table S1 in Supporting Information File 1). The second analysis involved an automatic feature selection method in which the molecular descriptors that were used had a nonzero rank according to the ChiSquaredAttributeEval method in Weka (see details in

Nanotechnology in the real world: Redeveloping the nanomaterial consumer products inventory

• Marina E. Vance,
• Todd Kuiken,
• Eric P. Vejerano,
• Sean P. McGinnis,
• Michael F. Hochella Jr.,
• David Rejeski and
• Matthew S. Hull

Beilstein J. Nanotechnol. 2015, 6, 1769–1780, doi:10.3762/bjnano.6.181

• eight new descriptors for consumer products, including information pertaining to the nanomaterials contained in each product. The project was motivated by the recognition that a diverse group of stakeholders from academia, industry, and state/federal government had become highly dependent on the

• nanotechnology. This effort resulted in archiving 316 products in the Health and Fitness category – mainly in the Personal Care and Clothing subcategories – with 86 and 78 products archived between 2012 and 2014, respectively. New nanomaterial descriptors Eight new product descriptors were introduced to

• the product, potential exposure pathways, “how much we know”, “researchers say”. The experimental section of this paper describes all new product descriptors. The results of the five new quantitative descriptors are presented and discussed below. Since the “nanomaterial shape and size”, “coating and

The eNanoMapper database for nanomaterial safety information

• Nina Jeliazkova,
• Charalampos Chomenidis,
• Philip Doganis,
• Bengt Fadeel,
• Roland Grafström,
• Barry Hardy,
• Janna Hastings,
• Markus Hegi,
• Vedrin Jeliazkov,
• Nikolay Kochev,
• Pekka Kohonen,
• Cristian R. Munteanu,
• Haralambos Sarimveis,
• Bart Smeets,
• Pantelis Sopasakis,
• Georgia Tsiliki,
• David Vorgrimmler and
• Egon Willighagen

Beilstein J. Nanotechnol. 2015, 6, 1609–1634, doi:10.3762/bjnano.6.165

• descriptors of engineered nanoparticles (mainly metal-based) and their potential toxicity. This dataset nicely demonstrates the complexity of the nanosafety domain. The ModNanoTox database provides physicochemical descriptors and toxic activities of nanoparticles from several studies. The database version

• from August 2013 includes 86 assays with more than 100 different endpoints affecting 45 species. Unfortunately, only a few nanoparticles (usually fewer than three) have been tested for each endpoint. Physicochemical descriptors for the characterisation of nanoparticles are incomplete as well (about 75

• cases the number of measured nanoparticle properties was very low. Most studies report only two to four different nanoparticle properties (descriptors) and the descriptor types are very inconsistent (overall 36 different descriptors, which results in very sparse matrices with a high number of missing