Repositório Comunidade: ECUM
https://hdl.handle.net/1822/24165
ECUM2024-03-29T11:02:21ZDesenvolvimento in silico de novos agentes antimicrobianos derivados da polimixina B
https://hdl.handle.net/1822/90242
Título: Desenvolvimento in silico de novos agentes antimicrobianos derivados da polimixina B
Autor: Inácio, João Miguel
Resumo: A resistência antimicrobiana (RAM) é um dos principais problemas de saúde pública da
atualidade, provocando uma morbimortalidade significativa. A evolução deste problema, agravada pelo
lento desenvolvimento de novos antimicrobianos, levou à reconsideração do uso das polimixinas,
fármacos que já estavam em desuso devido a sua alta toxicidade. Na tentativa de diminuir a sua
toxicidade e/ou melhorar a sua atividade antimicrobiana, vários análogos de polimixinas são gerados
através de diferentes estratégias, principalmente experimentais. Como tal, estão em falta abordagens
mais rápidas e fiáveis para tornar o design de análogos mais eficaz, a fim de combater a RAM o mais
rápido possível. A solução para acelerar a descoberta de novos fármacos provavelmente está no uso de
abordagens in silico, com métodos de machine learning (ML), devido ao seu ritmo mais rápido e baixo
custo.
Neste trabalho, a atividade de análogos da polimixina B foi modelada usando modelos semi quantitativos de relação estrutura-actividade baseados em ML. Neste contexto, foram aplicados três
algoritmos diferentes de ML (árvore de decisão, floresta aleatória e AdaBoost) em dez famílias diferentes
de descritores moleculares ao conjunto de dados de 413 pares molécula/microrganismo proveniente da
PubChem e dos ensaios laboratoriais.
O modelo DT/Estate_VSA destacou-se como promissor, com exatidões e previsões verdadeiras
altas, bem como previsões falsas negativas e falsas positivas muito baixas. Este modelo foi aplicado para
prever a atividade antimicrobiana de seis análogos das polimixinas B e E, sendo que todos são previstos
como promissores para Pseudomonas e não promissores para Acinetobacter. Para Escherichia, os três
análogos mais hidrofílicos foram previstos como promissores e os outros três como não promissores.
Estes análogos estão a ser sintetizados e posteriormente serão testados quanto a sua atividade in vitro.; Antimicrobial resistance (AMR) is one of the main public health problems today, causing
significant morbidity and mortality. The evolution of this problem, aggravated by the slow development of
new antimicrobials, led to the reconsideration of the use of polymyxins, drugs that were already in disuse
due to their high toxicity. In an attempt to decrease its toxicity and/or improve its antimicrobial activity,
several polymyxin analogues are generated through different strategies, mainly experimental. As such,
faster and more reliable approaches to make analogue design more effective in order to tackle AMR as
quickly as possible are lacking. The solution to accelerate the discovery of new drugs probably lies in the
use of in silico approaches, with machine learning (ML) methods, due to their faster pace and low cost.
In this work, the activity of polymyxin B analogues was modelled using semi-quantitative structure activity relationship models based on ML. In this context, three different ML algorithms (decision tree,
random forest, and AdaBoost) were applied in ten different families of molecular descriptors to the dataset
of 413 molecule/microorganism pairs from PubChem and laboratory assays.
The DT/Estate_VSA model stood out as promising, with high true accuracies and predictions, as
well as very low false negative and false positive predictions. This model was applied to predict the
antimicrobial activity of six polymyxin B and E analogues, all of which are predicted to be promising for
Pseudomonas and not promising for Acinetobacter. For Escherichia, the three most hydrophilic analogues
were predicted to be promising and the other three to be unpromising. These analogues are being
synthesized and will later be tested for their in vitro activity.
Descrição: Dissertação de mestrado em Química Medicinal
<b>Tipo</b>: masterThesis2024-03-28T11:31:47ZComplementos de optometria: protocolos práticos
https://hdl.handle.net/1822/90199
Título: Complementos de optometria: protocolos práticos
Autor: Jorge, Jorge
<b>Tipo</b>: pedagogicalPublicationSolving neural field equations using physics informed neural networks
https://hdl.handle.net/1822/90178
Título: Solving neural field equations using physics informed neural networks
Autor: Wojtak, Weronika; Bicho, Estela; Erlhagen, Wolfram
Resumo: This article presents an approach for solving neural field equations (NFEs) using Physics Informed Neural Networks
(PINNs). NFEs are integro-differential equations describing the spatio-temporal dynamics of neuronal populations in the cortex.
The traditional numerical methods for NFEs require significant computational effort due to the discretization of the spatial convolution.
The proposed approach leverages Fast Fourier Transforms (FFTs) to reduce the computational cost and improve efficiency. A
PINN, consisting of a surrogate network and a residual network, is trained to approximate the solutions of NFEs. The effectiveness
of the approach is demonstrated by solving the one-dimensional Amari equation, a commonly used neural field formulation. Our
results show that the accuracy of the PINN approach is comparable to traditional numerical methods. Future research directions include
optimizing hyperparameters, incorporating input terms in NFEs, exploring transfer learning, addressing the inverse problem,
and extending the approach to higher dimensions.
<b>Tipo</b>: conferencePaper2024-03-27T15:28:58ZQuinoline-based hydrazones for biocide detection: Machine learning-aided design of new TBT chemosensors
https://hdl.handle.net/1822/90133
Título: Quinoline-based hydrazones for biocide detection: Machine learning-aided design of new TBT chemosensors
Autor: Sousa, Rui Pedro Carvalho Lima; Teixeira, Filipe; Costa, Susana P. G.; Figueira, Rita Bacelar; Raposo, M. Manuela M.
Resumo: Antifouling compounds are used as paint components to mitigate biofouling on ships and submerged structures. One of the most known and used antifouling compounds is tributyltin (TBT). However, TBT is toxic to aquatic living beings, causing problems such as reduction of growth and imposex. The development of a TBT chemosensor could be of utter relevance in the building of an in-situ TBT monitoring device. Therefore, this work reports the synthesis of five new quinoline-based hydrazones (HZ) and two new quinoline-based thiosemicarbazones (TSC), with synthesis yields from 17 to 83 %. The compounds were tested in the presence of TBT, and some compounds of the group showed colorimetric or fluorimetric changes. The interaction between these compounds and TBT was tested by spectrophotometric or spectrofluorimetric titrations, which allowed to calculate the limit of detection (LOD) for each interaction. The fluorimetric interaction between HZ 4a and TBT was shown to be the most sensitive chemosensory method, with a LOD value of 1.7 µM.
A Ridge classifier model was developed to correlate the ability for TBT detection and the modification of the structure of each molecule. The validity of the proposed model was tested by assessing the TBT-sensing ability of the two novel TSC 5a and 5b, which were synthesized after the development of the model. These two compounds also showed colorimetric changes in the presence of TBT, with LODs of 13.8 and 3.1 µM, respectively, in good accordance with the model’s predictions. Further analysis of the model’s decision process provided some insights on the desirable properties of the novel quinoline-derived TBT optical chemosensors.
<b>Tipo</b>: article2024-03-27T10:38:41ZMiopía: teorías del desarrollo y técnicas de control
https://hdl.handle.net/1822/90090
Título: Miopía: teorías del desarrollo y técnicas de control
Autor: Jorge, Jorge
Resumo: Apontamentos em castelhano sobre "Miopía: teorías del desarrollo y técnicas de control"
<b>Tipo</b>: pedagogicalPublication