Please login first

List of accepted submissions

Show results per page
Find papers
  • Open access
  • 61 Reads
Machine learning-based prediction of toxicity of pesticide towards Americamysis bahia

Pesticides are toxic substances designed and widely applied throughout the world. However, their widespread use has received increasing attention from regulatory agencies due to the various acute and chronic effects they have on various organisms. In this study, QSTR (Quantitative Structure-Toxicity Relationship) models, based on nonlinear statistical techniques, have been established using five Machine Learning (ML) algorithms to predict the toxicity of pesticides on mysid shrimp (Americamysis bahia). The optimal nonlinear model (Random Forest, R2 = 0.983) was verified by internal (leave one cross-validation) and external validations. The validation results (qint2 = 0.815 and qext2 = 0.81) were satisfactory in predicting acute toxicity in the saltwater crustacean (A. bahia) compared to other models reported in the literature. In addition, this model also predicted the toxicity of some pesticides without experimental data. With a p(LC50) value of 12.102, bromadiolone was the most toxic compound. It is the active compound of the product RASTOP BLOCKS (rodenticide), which is classified in the category "Extremely Hazardous".

  • Open access
  • 59 Reads
Significance of the fourth atom bond connectivity index in predicting the physicochemical properties of polycyclic aromatic hydrocarbons

Chemical graph theory mainly deals with quantitative structure-activity (QSAR) and structure property relationships (QSPR) studies. A large number of topological indices have been introduced by various eminent researchers. These are found to be useful in those studies. In this paper, we focused on the fourth atom bond connectivity index and how it can be used for the accurate prediction of physicochemical properties of polycyclic aromatic hydrocarbons, considered as graphene fragments. These properties were boiling point, molar entropy, acentric factor, logarithm of octanol-water partition coefficient, and Kovats retention index.

  • Open access
  • 15 Reads
A probability problem suitable for Problem-Based Learning

We propose a simple probability problem for undergraduate level. This problem involves different branches of Mathematics, such as Graph Theory, Linear Algebra or hypergeometric sums, hence it is quite suitable to be used as Problem-Based Learning. In addition, the problem allows several variations so that it may be proposed to different groups of students at the same time. Finally, we propose that the students interpret the result obtained as combinations with repetition.

  • Open access
  • 5 Reads
Effect of antibiotics on gut microbiota in patients with cardiac surgery
, , , , , ,

Surgical site infection is a common complication after surgery. It is necessary to prevent or treat postoperative infection with antibiotics. The use of antibiotics could disturb the composition and/or functions of gut microbiota. The purpose of the present study was to explore the effect of antibiotics on gut microbiota in patients who underwent cardiac surgery. A total of 622 fecal samples were collected from 244 cardiac surgery patients. The V3–V4 hypervariable region of the bacterial 16S rRNA gene was amplified and sequenced on a MiSeq PE300. The gut microbiota diversity of samples was analyzed from alpha diversity at the OTU level and the distribution proportion of dominant bacteria at the genus level by a Circos graph, and the differences among different groups were compared by partial least squares discriminant analysis at the OTU level. As expected, antibiotics could perturb the composition of the gut microbiota in cardiac surgery patients. When antibiotics were administered over 7 days, the composition of the gut microbiota was significantly disturbed. The gut microbiota composition of patients returning to preantibiotic levels might need antibiotic withdrawal for at least 28 days.

  • Open access
  • 18 Reads
The future of AI in the EU: a preliminary analysis of the new proposal for a regulation

Rationale: protection of fundamental rights in the face of threats and risks linked to the development of AI tools/Strengthening innovation
Horizontal regulatory framework – not limited to specific sectors – proportional response to risk. Concept of AI: broad definition, any software that is developed using one or more of the techniques and strategies listed in Annex I and that can, for a given set of objectives defined by human beings, generate output information such as content, predictions, recommendations or decisions that influence the environments with which it interacts (Art. 3. I). Annex I techniques and strategies: Machine learning strategies, including supervised, unsupervised and reinforcement learning, which employ a wide variety of methods, including deep learning. Strategies based on logic and knowledge, especially the representation of knowledge, inductive programming (logic), knowledge bases, inference and deduction engines, expert and (symbolic) reasoning systems. Statistical strategies, Bayesian estimation, search methods and optimization

  • Open access
  • 10 Reads
Predictive algorithms and the use of automated decision systems for legal issues and regulatory affairs.

On this presentation is disucussed the basic definition of an algorithm as: a set of step-by-step instructions for solving a problem or performing a task. Examples are also given: Building blocks; Retrosynthesis algorithms; Natural computing and Bioinspired algorithms (Swarm intelligence (ant colony, AIS systems, genetic)); Fuzzy logic; Quantum computing algorithms. It is also discussed the legislation for their design. It is necessary to certify that in its preparation no biases have been voluntarily or involuntarily introduced that alter its operation according to the interests or inclinations of the programmers the accuracy of its predictions will depend on the quality of the data used in its preparation. European Parliament, it will be necessary to "regularly assess the representativeness of datasets [as well as] examine the accuracy and importance of predictions". Regulate the mechanisms that will ensure an adequate audit of artificial intelligence systems. Specific rules: REACH (Registration, Evaluation, Authorization and Restriction of Chemicals)) Regulation EC Nº 1907/2006 (art.13; 25). In addition, is discussed the use and Protection of Personal Data in the context of algorithms development and software programming. Protection of natural persons in relation to their personal data. Fundamental right (national level Art. 18 EC, Art. 1 LOPD at European level TFE Art.16, CDFUE Art. 8, Art. 2 GDPR) balance with other fundamental rights, in accordance with the principle of proportionality. Respect for personal, family, home and communications privacy; freedom of thought; consciousness and religion; freedom of expression; etc. Special categories of data: Sensitive data Art. 4 GDPR

  • Open access
  • 33 Reads
Contribution of vitamin K and gut microbiota to individual variability of warfarin in cardiac surgery patients
, , , , , ,

Warfarin is a commonly prescribed anticoagulant in clinic. It has large individual variability. The pharmacokinetics and pharmacodynamics (PKPD) of warfarin had been established. There also exist some unknown factors that could affect warfarin effect. The purpose of the study was to explore the contribution of vitamin K and gut microbiota to individual variability of warfarin in cardiac surgery patients. A total of 246 patients were enrolled in the present study. Serum and fecal samples were collected used to detect warfarin and vitamin K (VK1 and MK4) concentrations and the diversity of gut microbiota, respectively. The demographic characteristics, drug history, and CYP2C9 and VKORC1 genotype were recorded from the patients’ medical records. The INR and warfarin concentration bias were predicted according to the PKPD model. The results of INR and warfarin concentration predicted bias by the PKPD model were: 78.7% of Ideal prediction for INR, 66.7% of Ideal prediction for S-warfarin, 75.0% of Ideal prediction for R-warfarin. Only INR PE was compared with VK concentration and gut microbiota due to INR was the main indicator for warfarin therapy in clinic. The INR of patients decreased with increasing of VK concentration. The pharmacodynamic parameter C50 of S-warfarin, which was derived from the PKPD model, increased with increasing of VK concentration. The results diversity of gut microbiota showed that Prevotella might be associated with warfarin anticoagulation. In conclusion, vitamin K had some effect on the individual variability of warfarin. The gut microbiota might also have a certain effect on the variability of warfarin.

  • Open access
  • 9 Reads
New antitumor Ru-based compound derivatives optimized using in silico methods

Cancer has become one of leading causes of death around the globe, with female breast cancer as one of
the most prevalent. Among the multiple types of breast cancer (BC) identified to date, the triple-negative
(TN) subtype (lacking expression of estrogen and progesterone receptors and human epidermal growth
factor receptor 2) is associated with higher aggressiveness and poor prognosis [1]. TNBC lacks targeted
therapies and presents heterogeneous responses to treatment with traditional cisplatin-like drugs, in part
due to the development of multidrug resistance (MDR). TM34 is a Ruthenium-based compound that has
been suggested to be a more efficient and selective therapy than cisplatin [2]. More recently, new
derivatives of TM34 have been developed with increased selectivity by adding peptide sequences that are
recognized by receptor proteins from the FGFR family [3].
The main goal of this work is to study the interaction of several TM34 derivatives with a membrane model
(POPC) and to calculate their membrane crossing energy profiles that can be used to estimate the
membrane permeability coefficients. We used Molecular Dynamics simulations coupled with an
Umbrella-sampling scheme to obtain the potential of mean force profiles, which allowed the calculation
of the membrane permeability using the inhomogeneous solubility-diffusion model [4].

Acknowledgements: We acknowledge Fundação para a Ciência e Tecnologia (FCT) for funding through projects
UIDB/00100/2020 (CQE), UIDB/04046/2020 & UIDP/04046/2020 (BioISI), and PTDC/QUI-QIN/0146/2020. T.S.
Morais and M. Machuqueiro thank the CEECIND 2017 Initiative for projects CEECIND/00630/2017 and
CEECIND/02300/2017, respectively (acknowledging FCT, as well as POPH and FSE-European Social Fund).

[1] Rakha EA, Ellis IO. Triple-negative/basal-like breast cancer: review. Pathology. 2009;41: 40–47.
[2] Lin K, Zhao Z-Z, Bo H-B, Hao X-J, Wang J-Q. Applications of Ruthenium Complex in Tumor Diagnosis and Therapy.
Front Pharmacol. 2018;9: 1323.

[3] Machado JF, Machuqueiro M, Marques F, Robalo MP, Piedade MFM, Garcia MH, et al. Novel “ruthenium
cyclopentadienyl”--peptide conjugate complexes against human FGFR (+) breast cancer. Dalton Trans J Inorg Chem.
2020;49: 5974–5987.
[4] Dickson CJ, Hornak V, Pearlstein RA, Duca JS. Structure-Kinetic Relationships of Passive Membrane Permeation
from Multiscale Modeling. J Am Chem Soc. 2017;139: 442–452

  • Open access
  • 150 Reads
PyBindE: Development of a Simple Python MM-PBSA Implementation for Estimating Protein-Protein and Protein-Ligand Binding Energies

There are several approaches for calculating binding free energies, with single-trajectory MM-PBSA being particularly useful when the relative energy differences between configurations are most significant. These methods also become a very popular option since they can be applied to a vast variety of systems, including protein-protein, protein-ligand and even protein-membrane binding events. MM-PBSA can generate binding energies over time, with various force-fields, and can be used to investigate the impact of protonation changes in a complex stability.

With this in mind, we have just developed PyBindE, a single-trajectory MM-PBSA Python implementation designed to be easily inserted into existing MD protocols [1]. Although PyBindE is in its early stages of validation it has already been applied to a few different systems of protein-protein and protein-ligand. Here, we provide a detailed description of the PyBindE implementation, how it can be easily installed and inserted into MD simulations pipelines and some of the results from on-going and published projects [2].

[1] PyBindE: Molecular Mechanics Poisson-Boltzmann Surface Area (MMPBSA) calculations in protein-protein and
protein-ligand systems. Github; Available:
[2] Oliveira NFB, Rodrigues FEP, Vitorino JNM, Loureiro RJS, Faísca PFN, Machuqueiro M. Predicting stable binding
modes from simulated dimers of the D76N mutant of β 2-microglobulin. Comput Struct Biotechnol J. 2021;19: 5160–

  • Open access
  • 45 Reads
HTVS protocol to identify non-covalent inhibitors of CRM1

The protein function depends on its subcellular localization, as it determines the access to binding partners and enzymes that catalyze post-translational modifications. The best-studied export protein is the Chromosome Region Maintenance 1 (CRM1, also known as XPO1 or exportin 1), which is a transversal protein across all eukaryotic cells. Inhibition of CRM1 has long been idealized for the treatment of cancer and several viruses and it consists of binding a compound to the NES-binding groove to prevent the association of CRM1 with its cargo. However, all known inhibitors of CRM1 establish a covalent bond with Cys528, leading to high toxicities and impairing its in vivo application.

Until recently, all known inhibitors bound covalently to the NES-binding groove. However, in a recent paper, Lei et al. presented the first inhibitor that was able to bind non-covalently to the NES-binding groove - the non-covalent CRM1 inhibitor 1 (NCI-1) [1]. Unfortunately, and despite the name suggesting otherwise, this inhibitor also binds covalently to Cys528 in the wild-type form. Nevertheless, NCI-1 ability to bind to CRM1 non-covalently serves as a proof-of-concept that such inhibitors are viable and can be developed.

With the intent of discovering non-covalent inhibitors of CRM1, a high-throughput virtual screen (HTVS) protocol was developed and implemented using a database provided by our collaborator, Prof. Romano Silvestri (Head of Medicinal Chemistry, Sapienza Univ., Italy). This HTVS was done using both the NES-binding groove from the crystallographic structure available in the PDB (ID: 6TVO) and two new conformations sampled from MD simulations, which we expect to be better descriptors of the apo structure. The top rank compounds were selected and are now being tested experimentally by our collaborator, Professor Wolfgang Link (University of Madrid, Spain).

1. Lei Y, An Q, Shen X-F, Sui M, Li C, Jia D, et al. Structure-Guided Design of the First Noncovalent Small-Molecule Inhibitor of CRM1. J Med Chem. 2021;64: 6596–6607.