Antibody design using deep learning: from sequence and structure design to affinity maturation.

Deep learning has achieved impressive results in various fields such as computer vision and natural language processing, making it a powerful tool in biology. Its applications now encompass cellular image classification, genomic studies and drug discovery. While drug development traditionally focused deep learning applications on small molecules, recent innovations have incorporated it in the discovery and development of biological molecules, particularly antibodies. Researchers have devised novel techniques to streamline antibody development, combining in vitro and in silico methods. In particular, computational power expedites lead candidate generation, scaling and potential antibody development against complex antigens. This survey highlights significant advancements in protein design and optimization, specifically focusing on antibodies. This includes various aspects such as design, folding, antibody-antigen interactions docking and affinity maturation.

Exploiting the structure of biochemical pathways to investigate dynamical properties with neural networks for graphs.

MOTIVATION: Dynamical properties of biochemical pathways (BPs) help in understanding the functioning of living cells. Their in silico assessment requires simulating a dynamical system with a large number of parameters such as kinetic constants and species concentrations. Such simulations are based on numerical methods that can be time-expensive for large BPs. Moreover, parameters are often unknown and need to be estimated. RESULTS: We developed a framework for the prediction of dynamical properties of BPs directly from the structure of their graph representation. We represent BPs as Petri nets, which can be automatically generated, for instance, from standard SBML representations. The core of the framework is a neural network for graphs that extracts relevant information directly from the Petri net structure and exploits them to learn the association with the desired dynamical property. We show experimentally that the proposed approach reliably predicts a range of diverse dynamical properties (robustness, monotonicity, and sensitivity) while being faster than numerical methods at prediction time. In synergy with the neural network models, we propose a methodology based on Petri nets arc knock-out that allows the role of each molecule in the occurrence of a certain dynamical property to be better elucidated. The methodology also provides insights useful for interpreting the predictions made by the model. The results support the conjecture often considered in the context of systems biology that the BP structure plays a primary role in the assessment of its dynamical properties. AVAILABILITY AND IMPLEMENTATION: https://github.com/marcopodda/petri-bio (code), https://zenodo.org/record/7610382 (data).

Bayesian Sigmoid-Type Time Series Forecasting with Missing Data for Greenhouse Crops.

This paper follows an integrated approach of Internet of Things based sensing and machine learning for crop growth prediction in agriculture. A Dynamic Bayesian Network (DBN) relates crop growth associated measurement data to environmental control data via hidden states. The measurement data, having (non-linear) sigmoid-type dynamics, are instances of the two classes observed and missing, respectively. Considering that the time series of the logistic sigmoid function is the solution to a reciprocal linear dynamic model, the exact expectation-maximization algorithm can be applied to infer the hidden states and to learn the parameters of the model. At iterative convergence, the parameter estimates are then used to derive a predictor of the measurement data several days ahead. To evaluate the performance of the proposed DBN, we followed three cultivation cycles of micro-tomatoes (MicroTom) in a mini-greenhouse. The environmental parameters were temperature, converted into Growing Degree Days (GDD), and the solar irradiance, both at a daily granularity. The measurement data were Leaf Area Index (LAI) and Evapotranspiration (ET). Although measurement data were only available scarcely, it turned out that high quality measurement data predictions were possible up to three weeks ahead.

Mathematical modeling of drug resistance due to KRAS mutation in colorectal cancer.

The most challenging task in colorectal cancer research nowadays is to understand the development of acquired resistance to anti-EGFR drugs. The key reason for this problem is the KRAS mutations appearance after the treatment with monoclonal antibodies (moAb). Here we present a mathematical model for the analysis of KRAS mutations behavior in colorectal cancer with respect to moAb treatments. To evaluate the drug performance we have developed equations for two types of tumors cells, KRAS mutated and KRAS wild-type. Both tumor cell populations were treated with a combination of moAb and chemotherapy drugs. It was observed that even the minimal initial concentration of KRAS mutation before the treatment has the ability to make the tumor refractory to the treatment. Minor population of KRAS mutations has strong influence on large number of wild-type cells as well rendering them resistant to chemotherapy. Patient׳s immune responses are specifically taken into considerations and it is found that, in case of KRAS mutations, the immune strength does not affect medication efficacy. Finally, cetuximab (moAb) and irinotecan (chemotherapy) drugs are analyzed as first-line treatment of colorectal cancer with few KRAS mutated cells. Results show that this combined treatment could be only effective for patients with high immune strengths and it should not be recommended as first-line therapy for patients with moderate immune strengths or weak immune systems because of a potential risk of relapse, with KRAS mutant cells acquired resistance involved with them.

The role of deleterious mutations in the stability of hybridogenetic water frog complexes.

BACKGROUND: Some species of water frogs originated from hybridization between different species. Such hybrid populations have a particular reproduction system called hybridogenesis. In this paper we consider the two species Pelophylax ridibundus and Pelophylax lessonae, and their hybrids Pelophylax esculentus. P. lessonae and P. esculentus form stable complexes (L-E complexes) in which P. esculentus are hemiclonal. In L-E complexes all the transmitted genomes by P. esculentus carry deleterious mutations which are lethal in homozygosity. RESULTS: We analyze, by means of an individual based computational model, L-E complexes. The results of simulations based on the model show that, by eliminating deleterious mutations, L-E complexes collapse. In addition, simulations show that particular female preferences can contribute to the diffusion of deleterious mutations among all P. esculentus frogs. Finally, simulations show how L-E complexes react to the introduction of translocated P. ridibundus. CONCLUSIONS: The conclusions are the following: (i) deleterious mutations (combined with sexual preferences) strongly contribute to the stability of L-E complexes; (ii) female sexual choice can contribute to the diffusion of deleterious mutations; and (iii) the introduction of P. ridibundus can destabilize L-E complexes.

Drug Mechanism: A bioinformatic update.

A drug Mechanism of Action (MoA) is a complex biological phenomenon that describes how a bioactive compound produces a pharmacological effect. The complete knowledge of MoA is fundamental to fully understanding the drug activity. Over the years, many experimental methods have been developed and a huge quantity of data has been produced. Nowadays, considering the increasing omics data availability and the improvement of the accessible computational resources, the study of a drug MoA is conducted by integrating experimental and bioinformatics approaches. The development of new in silico solutions for this type of analysis is continuously ongoing; herein, an updating review on such bioinformatic methods is presented. The methodologies cited are based on multi-omics data integration in biochemical networks and Machine Learning (ML). The multiple types of usable input data and the advantages and disadvantages of each method have been analyzed, with a focus on their applications. Three specific research areas (i.e. cancer drug development, antibiotics discovery, and drug repurposing) have been chosen for their importance in the drug discovery fields in which the study of drug MoA, through novel bioinformatics approaches, is particularly productive.

RettDb: the Rett syndrome omics database to navigate the Rett syndrome genomic landscape.

Rett syndrome (RTT) is a neurodevelopmental disorder occurring almost exclusively in females and leading to a variety of impairments and disabilities from mild to severe. In >95% cases, RTT is due to mutations in the X-linked gene MECP2, but the molecular mechanisms determining RTT are unknown at present, and the complexity of the system is challenging. To facilitate and provide guidance to the unraveling of those mechanisms, we developed a database resource for the visualization and analysis of the genomic landscape in the context of wild-type or mutated Mecp2 gene in the mouse model. Our resource allows for the exploration of differential dynamics of gene expression and the prediction of new potential MECP2 target genes to decipher the RTT disorder molecular mechanisms. Database URL: https://biomedinfo.di.unipi.it/rett-database/.

Population dynamics with a mixed type of sexual and asexual reproduction in a fluctuating environment.

BACKGROUND: Carassius gibelio, a cyprinid fish from Eurasia, has the ability to reproduce both sexually and asexually. This fish is also known as an invasive species which colonized almost all continental Europe, most likely originating from Asia and Eastern Europe. Populations of both sexually and asexually reproducing individuals exist in sympatry. In this study we try to elucidate the advantages of such a mixed type of reproduction. We investigate the dynamics of two sympatric populations with sexual and asexual reproduction in a periodically fluctuating environment. We define an individual-based computational model in which genotypes are represented by L loci, and the environment is composed of L resources for which the two populations compete. RESULTS: Our model demonstrates advantageous population dynamics where the optimal percentage of asexual reproduction depends on selection strength, on the number of selected loci and on the timescale of environmental fluctuations. We show that the sexual reproduction is necessary for "generating" fit genotypes, while the asexual reproduction is suitable for "amplifying" them. The simulations show that the optimal percentage of asexual reproduction increases with the length of the environment stability period and decrease with the strength of the selection and the number of loci. CONCLUSIONS: In this paper we addressed the advantages of a mixed type of sexual and asexual reproduction in a changing environment and explored the idea that a species that is able to adapt itself to environmental fluctuation can easily colonize a new habitat. Our results could provide a possible explanation for the rapid and efficient invasion of species with a variable ratio of sexual and asexual reproduction such as Carassius gibelio.

Tumour suppression by immune system through stochastic oscillations.

The well-known Kirschner-Panetta model for tumour-immune System interplay [Kirschner, D., Panetta, J.C., 1998. Modelling immunotherapy of the tumour-immune interaction. J. Math. Biol. 37 (3), 235-252] reproduces a number of features of this essential interaction, but it excludes the possibility of tumour suppression by the immune system in the absence of therapy. Here we present a hybrid-stochastic version of that model. In this new framework, we show that in reality the model is also able to reproduce the suppression, through stochastic extinction after the first spike of an oscillation.