imagine (Institute of molecular genetics and genetic engineering)
Not a member yet
3088 research outputs found
Sort by
Evaluation of weaning diets for sustainable indoor largemouth bass (Micropterus salmoides) larviculture
No full textThis research evaluated the suitability of commercially available larval feeds, Otohime B2 (OB2), Aller Infa (AI), and Aqua Start (AS), and one Experimental Feed (EF), for the weaning of largemouth bass (Micropterus salmoides), LMB. Feeds were presented with various ω3 fatty acid levels/bioavailability (high in OB2 and AI), fat percent (high in OB2 and AS), free amino acid and short peptide (FAA+SP) levels (high in OB2), and various soluble protein (SPR) levels (high in AS and EF). Fish were co-fed Artemia plus OB2 from the 19th to 22nd -day post-hatching (DPH), then Artemia in addition to one of the four above diets for seven days, with complete Artemia removal on the 30th DPH. Fish were sampled on the 32nd DPH. Morphometry, digestive enzyme activities, hormonal status, skeleton, muscle development, and potentially pathogenic Flavobacterium spp. levels were estimated. Survival was high (96% or more) in all the weaning regimes. Weaning on OB2 was linked to a fast fish growth rate (14.29%/day), while both, OB2 and AI, supported skeleton development. Weight gain correlated with total fat, ash levels, free amino acids, and short peptide levels in the diet. Larvae weaned on soluble protein-rich AS and EF were presented with the lowest fish weight gain and skeleton development, and lower growth of potentially pathogenic Flavobacterium spp. This research suggests that the weaning diets for largemouth bass should have a balanced protein content and quality while allowing for the inclusion of fewer marine ingredients.Related to supplementary material:[https://imagine.imgge.bg.ac.rs/handle/123456789/2758]Related to preanalyzed data: [https://imagine.imgge.bg.ac.rs/handle/123456789/2759
HSA-MIR-18A-5P AND HSA-MIR-135B-5P EXPRESSION ASSOCIATION WITH NEOADJUVANT CHEMORADIOTHERAPY RESPONSE IN LOCALLY ADVANCED RECTAL CANCER PATIENTS
No full textThe first line of therapy for patients with locally advanced rectal cancer (LARC) is neoadjuvant
chemoradiotherapy (nCRT), aimed at downsizing and downstaging the tumour before surgery.
A complete response to nCRT is achieved in less than 20% of patients, highlighting an urget
need to identify novel predictive biomarkers to avoid unnecessary treatments for those who will
not benefit from nCRT. Given that hsa-miR-18a-5p and hsa-miR-135b-5p are deregulated in
colorectal cancer, the current study investigated their expression as potential biomarkers for
predicting response to nCRT.
The study included 19 LARC patients treated with nCRT. RNA was isolated from tumour tissue
both before and after nCRT. The relative expression of hsa-miR-18a-5p and hsa-miR-135b-5p,
normalised to RNU6B, was determined by qRT-PCR.
Hsa-miR-18a-5p was significantly downregulated in tumour tissue after nCRT compared to
tissue before therapy (p=0.001). There were no differences in the expression of hsa-miR-135b-
5p before and after therapy (p=0.114). No significant correlation was observed between two
studied miRNAs in LARC patients before and after nCRT. Responders and non-responders to
nCRT did not differ in expression of analysed miRNAs. Based on ROC analysis, neither hsamiR-
18a-5p (AUC=0.794, 95% CI=0.580-1.000, p=0.184) nor hsa-miR-135b-5p (AUC=0.382,
95% CI=0.132-0.632, p=0.595) were identified as predictive biomarkers.
Hsa-miR-18a-5p and hsa-miR-135b-5p cannot be used to predict nCRT response in LARC
patients. The decreased expression of hsa-miR-18a-5p in tumour tissue after nCRT may indicate
the therapeutic potential of this miRNA to be altered as a target in LARC patients. Further studies
should be conducted in a larger group of patients.VII Congress of the Serbian Genetic Society Zlatibor; October 2 to 5, 2024
Biocompatible Carbon Dots/Polyurethane Composites as Potential Agents for Combating Bacterial Biofilms: N-Doped Carbon Quantum Dots/Polyurethane and Gamma Ray-Modified Graphene Quantum Dots/Polyurethane Composites
No full textBackground: Pathogen bacteria appear and survive on various surfaces made of steel or glass. The existence of these bacteria in different forms causes significant problems in healthcare facilities and society. Therefore, the surface engineering of highly potent antimicrobial coatings is highly important in the 21st century, a period that began with a series of epidemics. Methods: In this study, we prepared two types of photodynamic polyurethane-based composite films encapsulated by N-doped carbon quantum dots and graphene quantum dots irradiated by gamma rays at a dose of 50 kGy, respectively. Further, we investigated their structural, optical, antibacterial, antibiofouling and biocompatibility properties. Results: Nanoelectrical and nanomechanical microscopy measurements revealed deviations in the structure of these quantum dots and polyurethane films. The Young’s modulus of elasticity of the carbon and graphene quantum dots was several times lower than that for single-walled carbon nanotubes (SWCNTs) with chirality (6,5). The electrical properties of the carbon and graphene quantum dots were quite similar to those of the SWCNTs (6,5). The polyurethane films with carbon quantum dots were much more elastic and smoother than the films with graphene quantum dots. Antibacterial tests indicated excellent antibacterial activities of these films against a wide range of tested bacteria, whereas the antibiofouling activities of both composite films showed the best results against the Staphylococcus aureus and Escherichia coli biofilms. Biocompatibility studies showed that neither composite film exhibited any cytotoxicity or hemolysis. Conclusions: Obtained results indicate that these composite films could be used as antibacterial surfaces in the healthcare facilities
Understanding the natural activation mechanism of the CRISPR-Cas immune system in Escherichia coli: A Computational Modeling Perspective
No full textCRISPR-Cas systems protect bacteria from viruses by using spacers, viral DNA fragments
stored within the CRISPR array in the bacterial genome. These spacers are transcribed and
then processed into crRNAs, guiding Cas proteins to eliminate complementary viral DNA
sequences. Despite CRISPR-Cas’s extensive use in biotechnology, its natural function in
bacteria, especially Escherichia coli, is not fully understood. In E. coli, CRISPR-Cas activity
is silenced by cooperatively bound H-NS proteins to the cas genes promoter. Viral DNA
with higher AT content might sequester some H-NS, alleviating this repression. The
transcriptional regulator LeuO, whose transcription is activated by BglJ-RcsB, can further
activate cas genes transcription.
This study explores whether a slight reduction in H-NS levels can provide initial
transcription derepression, and trigger a positive feedback loop activating the CRISPRCas
system expression through BglJ-RcsB and LeuO, leading to rapid crRNA generation
for defense against fast-replicating bacteriophages. We developed a dynamical model for
crRNA expression upon infection by foreign DNA and used the Random Forest machine
learning technique to identify parameters essential for achieving a sufficient crRNA
increase within 30 minutes of foreign DNA entry.
A bioinformatics analysis of 16,388 viruses and their host bacteria revealed a consistent,
slight increase in viral genomic AT content compared to hosts, suggesting reduced H-NS
levels available for repression upon bacteriophage infection. Significant reductions in H-NS
levels can rapidly induce crRNA expression, while smaller reductions require high H-NS
binding cooperativity and the baseline cellular H-NS level near to the H-NS equilibrium
binding constant.
Our findings indicate that CRISPR-Cas can quickly respond to fast-replicating
bacteriophages. The kinetic features, which are essential for understanding the natural
function of CRISPR-Cas and enhancing biotechnological applications, should inform
future experimental work.Book of abstracts: 5th Belgrade Bioinformatics Conference, Serbia, Belgrade,17-20 june 2024
Graphlet-based higher-order network embeddings: the past, the present and the future
No full textAt a high level, there exist two approaches for mining networks. Neighbourhood-based
approaches uncover groups (i.e., clusters) of tightly connected neighbouring nodes in a
network and make predictions based on guilt by association: two nodes are assumed
to be more likely to interact or share attributes if they belong to the same group(s) in
the network. Topology (i.e., structure) based approaches make predictions based on
structural similarity. The state-of-the-art methods to quantify local topology are based
on graphlets, which are small connected non-isomorphic induced subgraphs. To combine
neighbourhood and graphlet-based information, we defined graphlet adjacency, which
weighs the adjacency of two nodes based on their co-occurrence frequency on a given
graphlet (i.e., there is one type of adjacency for each graphlet). In this talk, we provide
an overview of various methodologies we generalised using graphlet adjacency, including
graphlet spectral embedding, graphlet eigencentrality, graphlet diffusion and hyperbolic
graphlet coalescent embedding, and show how we applied them to better describe the
functional organisation of various molecular networks and to better capture disease
mechanisms. Recently, we used graphlet based symmetries to improve random walk
based approaches. We conclude by presenting future research directions for new graphlet
adjacency-based methods and applications.Book of abstracts: 5th Belgrade Bioinformatics Conference, Serbia, Belgrade,17-20 june 2024
The genetic pathways of ferroptosis-related processes in multiple sclerosis
No full textMultiple sclerosis (MS), a chronic inflammatory and neurodegenerative disease with no
current cure, in its aetiology comprehend: susceptibility of central nervous system (CNS) to
oxidative damage, mitochondrial dysfunction, impaired iron metabolism, which all lead to
ferroptosis.
Therapeutic capacities to modulate ferroptosis, recently discovered cell death, have been
highlighted, in vitro, and require further both bionformatic and experimental research to
complement lack of studies in human neurodegenerative diseases. By investigating entire
transcriptome in MS patients, we have identified enrichment of the Ferroptosis pathway in
DEGs, before clear experimental evidence of its role in MS were presented. Consequently,
the FerroReg project aimed to further investigate transcriptional and post-transcriptional
regulation of ferroptosis related processes in MS. There is still a lack of specific molecular/
genetic markers that reflect ferroptosis-related molecular changes. A curated assemblage
of the ferroptosis-related genes has been performed to create custom panel of 138 genes
for targeted mRNA sequencing. The genes were classified according to their roles in relevant
processes: lipid oxidative metabolism, antioxidant defence and iron metabolism, next to
their proposed direct/indirect effect on ferroptosis. Additionally, 14 encoded transcription
regulators which were associated with ferroptosis were included. Ferroptosis draws attention
as a biological pathway that could be modulated, to achieve reduction of both inflammation
and neurodegeneration in the central nervous system. Accordingly, gene expression was
analysed with regard to disease severity, taking into account the disease modifying therapy.
Applied approach overcomes the limitation of previous bioinformatic studies, which lacked
the clinical data in existing gene expression data sets. Among identified DEGs, 18 genes
were upregulated while 8 genes were downregulated in progressive patients compared to
mild phenotype. The enrichment analysis performed on the minimum network confirmed
the strong enrichment of the ferroptosis pathway while two DEGs were classified as hub
molecules: TP53 and CDKN1A.Book of abstracts: 5th Belgrade Bioinformatics Conference, Serbia, Belgrade,17-20 june 2024
Spectral Clustering for Transcriptomics Data: an Approximate Column Sampling Approach on the GPU
No full textClustering is one of the central methods in bioinformatics processing pipelines. It is used
to capture hidden patterns in high-dimensional data, especially at the genomic level
where large quantities of gene expression data have been produced with the advent of
high-throughput sequencing technologies. Clustering reveals natural structure in the
data, helping in understanding gene function, regulations, and cellular processes, cell
assignment and subtyping, and other downstream analyses.
Spectral clustering has been successfully used to discover structure and patterns in data
for bioinformatics research in the past with short execution times and high accuracy.
However, it can be computationally expensive for large-scale datasets that are present
in contemporary single-cell RNA (sc-RNA) and spatial transcriptomics (ST) applications.
Numerous approximate spectral clustering algorithms have been proposed in the open
literature to improve efficiency and scalability, while maintaining clustering quality.
Spectral clustering uses linear algebra operations which can be efficiently implemented on
modern central processing units (CPUs) and graphics processing units (GPUs). In our work,
we implemented an approximate, parallel spectral clustering method based on column
sampling and the Nystrom method on the GPU. Our implementation significantly reduces
both the computational and memory requirements of the previous methods, enabling the
method to handle datasets of more than 106 samples and thousands of features.
We evaluated our approach using general datasets containing images of handwritten
digits, as well as several datasets from single-cell and spatial transcriptomics sequencing
of mouse brain. Datasets from sc-RNA and ST domains were characterized by a modestly
large number of points and a relatively higher number of clusters compared to the general
datasets. We compared our solution to the widely used Leiden algorithm in terms of
speedup, ARI and NMI score. Our results showed significant time advantage and scalability
over Leiden algorithm of up to a hundred times, albeit with to some extent lower ARI and
NMI scores.Book of abstracts: 5th Belgrade Bioinformatics Conference, Serbia, Belgrade,17-20 june 202
Detecting Genetic Interactions with Visible Neural Networks
No full textNon-linear interactions among single nucleotide polymorphisms (SNPs), genes, and pathways
play an important role in human diseases, but identifying these interactions is a challenging
task. Neural networks are state-of-the-art predictors in many domains due to their ability to
analyze big data and model complex patterns, including non-linear interactions. In genetics,
visible neural networks are gaining popularity as they provide insight into the most important
SNPs, genes and pathways for prediction. Visible neural networks use prior knowledge (e.g.
gene and pathway annotations) to define the connections between nodes in the network,
making them sparse and interpretable. Currently, most of these networks provide measures
for the importance of SNPs, genes, and pathways but lack details on the nature of the
interactions. Here, we explore different methods to detect non-linear interactions with
visible neural networks. We adapt and speed up existing methods, create a comprehensive
benchmark with simulated data from GAMETES and EpiGEN, and demonstrate that these
methods can extract multiple types of interactions from trained visible neural networks. We
also highlight the strengths and weaknesses of the various methods in different settings,
providing guidelines for general use-cases. Finally, we apply these methods to a genomewide
case-control study of inflammatory bowel disease and find high consistency of epistasis
signals. Follow-up association testing revealed seven statistically significant epistasis SNP
pairs. The results and the code to reproduce the analysis are available at https://github.com/
ArnovanHilten/GenNet .Book of abstracts: 5th Belgrade Bioinformatics Conference, Serbia, Belgrade,17-20 june 2024
Transgenerational transmission of post-zygotic mutations suggests symmetric contribution of first two blastomeres to human germline
No full textLittle is known about the origin of germ cells in humans. We previously leveraged post-zygotic
mutations to reconstruct zygote-rooted cell lineage ancestry trees in a phenotypically normal
woman, termed NC0. Here, by sequencing the genome of her children and their father, we
analyzed the transmission of early pre-gastrulation lineages and corresponding mutations
across human generations. We found that the germline in NC0 is polyclonal and is founded
by at least two cells likely descending from the two blastomeres arising from the first zygotic
cleavage. Analyses of public data from several multi-children families and from 1,934 familial
quads confirmed this finding in larger cohorts, revealing that known imbalances of up to 90:10
in early lineages allocation in somatic tissues are not reflected in transmission to offspring,
establishing a fundamental difference in lineage allocation between the soma and the
germline. Analyses of all the data consistently suggest that germline has a balanced 50:50
lineage allocation from the first two blastomeres.Book of abstracts: 5th Belgrade Bioinformatics Conference, Serbia, Belgrade,17-20 june 2024
Advances in a comprehensive understanding of alpha-1 antitrypsin glycosylation-data mining within recent publications
No full textAlpha-1 antitrypsin (AAT) is the serin protease archetype, with biological roles shedding
from the neutrophil-derived protease activities control to a comprehensive innate
immunity modulation via interactions with the numerous proteins, cytokines, and
cells. UniProt Knowledgebase® reports three glycosylation sites for AAT, with attached
N-glycans contributing to approximately 12% of the total molecular mass and having a
crucial role in immunomodulatory activities. The PubMed database was searched for
“alpha-1 antitrypsin” and “glycosylation” using the following filters: full text available, the
publication date of 10 years, and the preprints excluded. The search retrieved 73 results,
and after the content analysis by the authors, 60 remained relevant, six reviews and 54
original research articles. Most studies used human samples (24 used serum/plasma, 23
cell cultures, four cerebrospinal or other non-standard sample fluids, and three studies
combined the different sample types). There were also two studies on animals and one on
plant models. Analysis of AAT glycoforms was the core of lab methodology in 52 studies,
while six studies analyzed the released glycans. Thirteen studies focused on analytical
protocols’ development or optimization. The glycosylation-related AAT structural and
functional properties were among the aims of 25 studies, while 10 papers assessed
these features from a glycoengineering perspective. The pathophysiological mechanisms
relating to AAT glycosylation features were studied in 10 papers, and 20 studies identified
AAT glycoforms as biomarker candidates, mainly in the oncology field (pancreatic, liver,
oral, and ovarian cancer). These pioneer data mining results indicate the availability of
comprehensive data about AAT glycosylation, thereby establishing a solid basis for further
scientific and innovation efforts