Machine Learning Strategies for Improved Phenotype Prediction in Underrepresented Populations.

Precision medicine models often perform better for populations of European ancestry due to the over-representation of this group in the genomic datasets and large-scale biobanks from which the models are constructed. As a result, prediction models may misrepresent or provide less accurate treatment recommendations for underrepresented populations, contributing to health disparities. This study introduces an adaptable machine learning toolkit that integrates multiple existing methodologies and novel techniques to enhance the prediction accuracy for underrepresented populations in genomic datasets. By leveraging machine learning techniques, including gradient boosting and automated methods, coupled with novel population-conditional re-sampling techniques, our method significantly improves the phenotypic prediction from single nucleotide polymorphism (SNP) data for diverse populations. We evaluate our approach using the UK Biobank, which is composed primarily of British individuals with European ancestry, and a minority representation of groups with Asian and African ancestry. Performance metrics demonstrate substantial improvements in phenotype prediction for underrepresented groups, achieving prediction accuracy comparable to that of the majority group. This approach represents a significant step towards improving prediction accuracy amidst current dataset diversity challenges. By integrating a tailored pipeline, our approach fosters more equitable validity and utility of statistical genetics methods, paving the way for more inclusive models and outcomes.

Machine Learning Strategies for Improved Phenotype Prediction in Underrepresented Populations.

Precision medicine models often perform better for populations of European ancestry due to the over-representation of this group in the genomic datasets and large-scale biobanks from which the models are constructed. As a result, prediction models may misrepresent or provide less accurate treatment recommendations for underrepresented populations, contributing to health disparities. This study introduces an adaptable machine learning toolkit that integrates multiple existing methodologies and novel techniques to enhance the prediction accuracy for underrepresented populations in genomic datasets. By leveraging machine learning techniques, including gradient boosting and automated methods, coupled with novel population-conditional re-sampling techniques, our method significantly improves the phenotypic prediction from single nucleotide polymorphism (SNP) data for diverse populations. We evaluate our approach using the UK Biobank, which is composed primarily of British individuals with European ancestry, and a minority representation of groups with Asian and African ancestry. Performance metrics demonstrate substantial improvements in phenotype prediction for underrepresented groups, achieving prediction accuracy comparable to that of the majority group. This approach represents a significant step towards improving prediction accuracy amidst current dataset diversity challenges. By integrating a tailored pipeline, our approach fosters more equitable validity and utility of statistical genetics methods, paving the way for more inclusive models and outcomes.

Age-differentiated incentives for adaptive behavior during epidemics produce oscillatory and chaotic dynamics.

Heterogeneity in contact patterns, mortality rates, and transmissibility among and between different age classes can have significant effects on epidemic outcomes. Adaptive behavior in response to the spread of an infectious pathogen may give rise to complex epidemiological dynamics. Here we model an infectious disease in which adaptive behavior incentives, and mortality rates, can vary between two and three age classes. The model indicates that age-dependent variability in infection aversion can produce more complex epidemic dynamics at lower levels of pathogen transmissibility and that those at less risk of infection can still drive complexity in the dynamics of those at higher risk of infection. Policymakers should consider the interdependence of such heterogeneous groups when making decisions.

Folylpolyglutamate synthetase mRNA G-quadruplexes regulate its cell protrusion localization and enhance a cancer cell invasive phenotype upon folate repletion.

BACKGROUND: Folates are crucial for the biosynthesis of nucleotides and amino acids, essential for cell proliferation and development. Folate deficiency induces DNA damage, developmental defects, and tumorigenicity. The obligatory enzyme folylpolyglutamate synthetase (FPGS) mediates intracellular folate retention via cytosolic and mitochondrial folate polyglutamylation. Our previous paper demonstrated the association of the cytosolic FPGS (cFPGS) with the cytoskeleton and various cell protrusion proteins. Based on these recent findings, the aim of the current study was to investigate the potential role of cFPGS at cell protrusions. RESULTS: Here we uncovered a central role for two G-quadruplex (GQ) motifs in the 3'UTR of FPGS mediating the localization of cFPGS mRNA and protein at cell protrusions. Using the MBSV6-loop reporter system and fluorescence microscopy, we demonstrate that following folate deprivation, cFPGS mRNA is retained in the endoplasmic reticulum, whereas upon 15 min of folate repletion, this mRNA is rapidly translocated to cell protrusions in a 3'UTR- and actin-dependent manner. The actin dependency of this folate-induced mRNA translocation is shown by treatment with Latrunculin B and inhibitors of the Ras homolog family member A (RhoA) pathway. Upon folate repletion, the FPGS 3'UTR GQs induce an amoeboid/mesenchymal hybrid cell phenotype during migration and invasion through a collagen gel matrix. Targeted disruption of the 3'UTR GQ motifs by introducing point mutations or masking them by antisense oligonucleotides abrogated cell protrusion targeting of cFPGS mRNA. CONCLUSIONS: Collectively, the GQ motifs within the 3'UTR of FPGS regulate its transcript and protein localization at cell protrusions in response to a folate cue, inducing cancer cell invasive phenotype. These novel findings suggest that the 3'UTR GQ motifs of FPGS constitute an attractive druggable target aimed at inhibition of cancer invasion and metastasis.

Deciphering molecular mechanisms underlying chemoresistance in relapsed AML patients: towards precision medicine overcoming drug resistance.

BACKGROUND: Acute myeloid leukemia (AML) remains a devastating disease with a 5-year survival rate of less than 30%. AML treatment has undergone significant changes in recent years, incorporating novel targeted therapies along with improvements in allogeneic bone marrow transplantation techniques. However, the standard of care remains cytarabine and anthracyclines, and the primary hindrance towards curative treatment is the frequent emergence of intrinsic and acquired anticancer drug resistance. In this respect, patients presenting with chemoresistant AML face dismal prognosis even with most advanced therapies. Herein, we aimed to explore the potential implementation of the characterization of chemoresistance mechanisms in individual AML patients towards efficacious personalized medicine. METHODS: Towards the identification of tailored treatments for individual patients, we herein present the cases of relapsed AML patients, and compare them to patients displaying durable remissions following the same chemotherapeutic induction treatment. We quantified the expression levels of specific genes mediating drug transport and metabolism, nucleotide biosynthesis, and apoptosis, in order to decipher the molecular mechanisms underlying intrinsic and/or acquired chemoresistance modalities in relapsed patients. This was achieved by real-time PCR using patient cDNA, and could be readily implemented in the clinical setting. RESULTS: This analysis revealed pre-existing differences in gene expression levels between the relapsed patients and patients with lasting remissions, as well as drug-induced alterations at different relapse stages compared to diagnosis. Each of the relapsed patients displayed unique chemoresistance mechanisms following similar treatment protocols, which could have been missed in a large study aimed at identifying common drug resistance determinants. CONCLUSIONS: Our findings emphasize the need for standardized evaluation of key drug transport and metabolism genes as an integral component of routine AML management, thereby allowing for the selection of treatments of choice for individual patients. This approach could facilitate the design of efficacious personalized treatment regimens, thereby reducing relapse rates of therapy refractory disease.

Surmounting Cytarabine-resistance in acute myeloblastic leukemia cells and specimens with a synergistic combination of hydroxyurea and azidothymidine.

Acute myeloid leukemia (AML) patients display dismal prognosis due to high prevalence of refractory and relapsed disease resulting from chemoresistance. Treatment protocols, primarily based on the anchor drug Cytarabine, remained chiefly unchanged in the past 50 years with no standardized salvage regimens. Herein we aimed at exploring potential pre-clinical treatment strategies to surmount Cytarabine resistance in human AML cells. We established Cytarabine-resistant sublines derived from human leukemia K562 and Kasumi cells, and characterized the expression of Cytarabine-related genes using real-time PCR and Western blot analyses to uncover the mechanisms underlying their Cytarabine resistance. This was followed by growth inhibition assays and isobologram analyses testing the sublines' sensitivity to the clinically approved drugs hydroxyurea (HU) and azidothymidine (AZT), compared to their parental cells. All Cytarabine-resistant sublines lost deoxycytidine kinase (dCK) expression, rendering them refractory to Cytarabine. Loss of dCK function involved dCK gene deletions and/or a novel frameshift mutation leading to dCK transcript degradation via nonsense-mediated decay. Cytarabine-resistant sublines displayed hypersensitivity to HU and AZT compared to parental cells; HU and AZT combinations exhibited a marked synergistic growth inhibition effect on leukemic cells, which was intensified upon acquisition of Cytarabine-resistance. In contrast, HU and AZT combination showed an antagonistic effect in non-malignant cells. Finally, HU and AZT synergism was demonstrated on peripheral blood specimens from AML patients. These findings identify a promising HU and AZT combination for the possible future treatment of relapsed and refractory AML, while sparing normal tissues from untoward toxicity.

The JmjN domain as a dimerization interface and a targeted inhibitor of KDM4 demethylase activity.

Histone methylation is regulated to shape the epigenome by modulating DNA compaction, thus playing central roles in fundamental chromatin-based processes including transcriptional regulation, DNA repair and cell proliferation. Histone methylation is erased by demethylases including the well-established KDM4 subfamily members, however, little is known about their dimerization capacity and its impact on their demethylase activity. Using the powerful bimolecular fluorescence complementation technique, we herein show the in situ formation of human KDM4A and KDM4C homodimers and heterodimers in nuclei of live transfectant cells and evaluate their H3K9me3 demethylation activity. Using size exclusion HPLC as well as Western blot analysis, we show that endogenous KDM4C undergoes dimerization under physiological conditions. Importantly, we identify the JmjN domain as the KDM4C dimerization interface and pin-point specific charged residues therein to be essential for this dimerization. We further demonstrate that KDM4A/C dimerization is absolutely required for their demethylase activity which was abolished by the expression of free JmjN peptides. In contrast, KDM4B does not dimerize and functions as a monomer, and hence was not affected by free JmjN expression. KDM4 proteins are overexpressed in numerous malignancies and their pharmacological inhibition or depletion in cancer cells was shown to impair tumor cell proliferation, invasion and metastasis. Thus, the KDM4 dimer-interactome emerging from the present study bears potential implications for cancer therapeutics via selective inhibition of KDM4A/C demethylase activity using JmjN-based peptidomimetics.