ABSTRACT
Mesomelic skeletal dysplasia is a heterogeneous group of skeletal disorders that has grown since the molecular basis of these conditions is in the process of research and discovery. Here, we report a Brazilian family with eight affected members over three generations with a phenotype similar to mesomelic Kantaputra dysplasia. This family presents marked shortening of the upper limbs with hypotrophy of the lower limbs and clubfeet without synostosis. Array-based CNV analysis and exome sequencing of four family members failed to show any region or gene candidate. Interestingly, males were more severely affected than females in this family, suggesting that gender differences could play a role in the phenotypic expressivity of this condition.
Subject(s)
Gonadal Dysgenesis , Osteochondrodysplasias , Male , Female , Humans , Sex Factors , Osteochondrodysplasias/genetics , Family , PhenotypeABSTRACT
Evolving in sync with the computation revolution over the past 30 years, computational biology has emerged as a mature scientific field. While the field has made major contributions toward improving scientific knowledge and human health, individual computational biology practitioners at various institutions often languish in career development. As optimistic biologists passionate about the future of our field, we propose solutions for both eager and reluctant individual scientists, institutions, publishers, funding agencies, and educators to fully embrace computational biology. We believe that in order to pave the way for the next generation of discoveries, we need to improve recognition for computational biologists and better align pathways of career success with pathways of scientific progress. With 10 outlined steps, we call on all adjacent fields to move away from the traditional individual, single-discipline investigator research model and embrace multidisciplinary, data-driven, team science.
Subject(s)
Computational Biology , Budgets , Cooperative Behavior , Humans , Interdisciplinary Research , Mentoring , Motivation , Publications , Reward , SoftwareABSTRACT
Mesial temporal lobe epilepsy (MTLE) is a chronic neurological disorder characterized by the occurrence of seizures, and histopathological abnormalities in the mesial temporal lobe structures, mainly hippocampal sclerosis (HS). We used a multi-omics approach to determine the profile of transcript and protein expression in the dorsal and ventral hippocampal dentate gyrus (DG) and Cornu Ammonis 3 (CA3) in an animal model of MTLE induced by pilocarpine. We performed label-free proteomics and RNAseq from laser-microdissected tissue isolated from pilocarpine-induced Wistar rats. We divided the DG and CA3 into dorsal and ventral areas and analyzed them separately. We performed a data integration analysis and evaluated enriched signaling pathways, as well as the integrated networks generated based on the gene ontology processes. Our results indicate differences in the transcriptomic and proteomic profiles among the DG and the CA3 subfields of the hippocampus. Moreover, our data suggest that epileptogenesis is enhanced in the CA3 region when compared to the DG, with most abnormalities in transcript and protein levels occurring in the CA3. Furthermore, our results show that the epileptogenesis in the pilocarpine model involves predominantly abnormal regulation of excitatory neuronal mechanisms mediated by N-methyl D-aspartate (NMDA) receptors, changes in the serotonin signaling, and neuronal activity controlled by calcium/calmodulin-dependent protein kinase (CaMK) regulation and leucine-rich repeat kinase 2 (LRRK2)/WNT signaling pathways.
Subject(s)
Epilepsy, Temporal Lobe , Animals , Epilepsy, Temporal Lobe/pathology , Hippocampus/metabolism , Pilocarpine/toxicity , Proteomics , Rats , Rats, WistarABSTRACT
Molecular diagnosis is important to provide accurate genetic counseling of skeletal dysplasias (SD). Although next-generation sequencing (NGS) techniques are currently the preferred methods for analyzing these conditions, some of the published results have not shown a detection rate as high as it would be expected. The present study aimed to assess the diagnostic yield of targeted NGS combined with Sanger sequencing (SS) for low-coverage exons of genes of interest and exome sequencing (ES) in a series of patients with rare SD and use two patients as an example of our strategy. This study used two different in-house panels. Of 93 variants found in 88/114 (77%) patients, 57 are novel. The pathogenic variants found in the following genes: B3GALT6, PCYT1A, INPPL1, LIFR, of four patients were only detected by SS. In conclusion, the high diagnostic yield reached in the present study can be attributed to both a good selection of patients and the utilization of the SS for the insufficiently covered regions. Additionally, the two case reports-a patient with acrodysostosis related to PRKAR1A and another with ciliopathy associated with KIAA0753, add new and relevant clinical information to the current knowledge.
Subject(s)
Dysostoses , Osteochondrodysplasias , Choline-Phosphate Cytidylyltransferase , Galactosyltransferases , Genetic Counseling , High-Throughput Nucleotide Sequencing , Humans , Exome SequencingABSTRACT
For a better interpretation of variants, evidence-based databases, such as ClinVar, compile data on the presumed relationships between variants and phenotypes. In this study, we aimed to analyze the pattern of sequencing depth in variants from whole-exome sequencing data in the 1000 Genomes project phase 3, focusing on the variants present in the ClinVar database that were predicted to affect protein-coding regions. We demonstrate that the distribution of the sequencing depth varies across different sequencing centers (pair-wise comparison, p < 0.001). Most importantly, we found that the distribution pattern of sequencing depth is specific to each facility, making it possible to correctly assign 96.9% of the samples to their sequencing center. Thus, indicating the presence of a systematic bias, related to the methods used in the different facilities, which generates significant variations in breadth and depth in whole-exome sequencing data in clinically relevant regions. Our results show that methodological differences, leading to significant heterogeneity in sequencing depth, may potentially influence the accuracy of genetic diagnosis. Furthermore, our findings highlight how it is still challenging to integrate results from different sequencing centers, which may also have an impact on genomic research.
ABSTRACT
OBJECTIVE: Focal cortical dysplasias (FCDs) are an important cause of drug-resistant epilepsy. In this work, we aimed to investigate whether abnormal gene regulation, mediated by microRNA, could be involved in FCD type II. METHODS: We used total RNA from the brain tissue of 16 patients with FCD type II and 28 controls. MicroRNA expression was initially assessed by microarray. Quantitative polymerase chain reaction, in situ hybridization, luciferase reporter assays, and deep sequencing for genes in the mTOR pathway were performed to validate and further explore our initial study. RESULTS: hsa-let-7f (p = 0.039), hsa-miR-31 (p = 0.0078), and hsa-miR34a (p = 0.021) were downregulated in FCD type II, whereas a transcription factor involved in neuronal and glial fate specification, NEUROG2 (p < 0.05), was upregulated. We also found that the RND2 gene, a NEUROG2-target, is upregulated (p < 0.001). In vitro experiments showed that hsa-miR-34a downregulates NEUROG2 by binding to its 5'-untranslated region. Moreover, we observed strong nuclear expression of NEUROG2 in balloon cells and dysmorphic neurons and found that 28.5% of our patients presented brain somatic mutations in genes of the mTOR pathway. INTERPRETATION: Our findings suggest a new molecular mechanism, in which NEUROG2 has a pivotal and central role in the pathogenesis of FCD type II. In this way, we found that the downregulation of hsa-miR-34a leads to upregulation of NEUROG2, and consequently to overexpression of the RND2 gene. These findings indicate that a faulty coupling in neuronal differentiation and migration mechanisms may explain the presence of aberrant cells and complete dyslamination in FCD type II. Ann Neurol 2018;83:623-635.
Subject(s)
Basic Helix-Loop-Helix Transcription Factors/metabolism , Epilepsy/metabolism , Focal Dermal Hypoplasia/metabolism , Malformations of Cortical Development/metabolism , Nerve Tissue Proteins/metabolism , Adolescent , Adult , Child , Child, Preschool , Drug Resistant Epilepsy/genetics , Epilepsy/drug therapy , Epilepsy/genetics , Female , Focal Dermal Hypoplasia/genetics , Humans , Infant , Male , Neurons/metabolism , TOR Serine-Threonine Kinases/metabolism , Transcription Factors/genetics , Young Adult , rho GTP-Binding Proteins/metabolismABSTRACT
Bioconductor is an open-source, open-development software project for the analysis and comprehension of high-throughput data in genomics and molecular biology. The project aims to enable interdisciplinary research, collaboration and rapid development of scientific software. Based on the statistical programming language R, Bioconductor comprises 934 interoperable packages contributed by a large, diverse community of scientists. Packages cover a range of bioinformatic and statistical applications. They undergo formal initial review and continuous automated testing. We present an overview for prospective users and contributors.
Subject(s)
Computational Biology , Gene Expression Profiling , Genomics/methods , High-Throughput Screening Assays/methods , Software , Programming Languages , User-Computer InterfaceABSTRACT
High-throughput technologies are widely used in the field of functional genomics and used in an increasing number of applications. For many 'wet lab' scientists, the analysis of the large amount of data generated by such technologies is a major bottleneck that can only be overcome through very specialized training in advanced data analysis methodologies and the use of dedicated bioinformatics software tools. In this article, we wish to discuss the challenges related to delivering training in the analysis of high-throughput sequencing data and how we addressed these challenges in the hands-on training courses that we have developed at the European Bioinformatics Institute.
Subject(s)
Computational Biology/education , Genomics/statistics & numerical data , High-Throughput Nucleotide Sequencing/statistics & numerical data , Academies and Institutes , Curriculum , Data Interpretation, Statistical , Europe , Faculty , Humans , Software , TeachingABSTRACT
Depression is a complex and multifactorial disease, affecting about 6.5% of the elderly population in what is referred to as late-life depression (LLD). Despite its public health relevance, there is still limited information about the molecular mechanisms of LLD. We analyzed the blood plasma of 50 older adults, 19 with LLD and 31 controls, through untargeted mass spectrometry, and used systems biology tools to identify biochemical pathways and biological processes dysregulated in the disease. We found 96 differentially expressed proteins between LLD patients and control individuals. Using elastic-net regression, we generated a panel of 75 proteins that comprises a potential model for determining the molecular signature of LLD. We also showed that biological pathways related to vesicle-mediated transport and voltage-dependent calcium channels may be dysregulated in LLD. These data can help to build an understanding of the molecular basis of LLD, offering an integrated view of the biomolecular alterations that occur in this disorder. SIGNIFICANCE: Major depressive disorder in the elderly, called late-life depression (LLD), is a common and disabling disorder, with recent prevalence estimates of 6.5% in the general population. Despite the public health relevance, there is still limited information about the molecular mechanisms of LLD. The findings in this paper shed light on LLD heterogeneous biological mechanisms. We uncovered a potential novel biomolecular signature for LLD and biological pathways related to this condition which can be targets for the development of novel interventions for prevention, early diagnosis, and treatment of LLD.
Subject(s)
Depressive Disorder, Major , Aged , Calcium Channels , Humans , Plasma , Proteins , ProteomicsABSTRACT
OBJECTIVES: We compared the proteomic signatures of the hippocampal lesion induced in three different animal models of mesial temporal lobe epilepsy with hippocampal sclerosis (MTLE+HS): the systemic pilocarpine model (PILO), the intracerebroventricular kainic acid model (KA), and the perforant pathway stimulation model (PPS). METHODS: We used shotgun proteomics to analyze the proteomes and find enriched biological pathways of the dorsal and ventral dentate gyrus (DG) isolated from the hippocampi of the three animal models. We also compared the proteomes obtained in the animal models to that from the DG of patients with pharmacoresistant MTLE+HS. RESULTS: We found that each animal model presents specific profiles of proteomic changes. The PILO model showed responses predominantly related to neuronal excitatory imbalance. The KA model revealed alterations mainly in synaptic activity. The PPS model displayed abnormalities in metabolism and oxidative stress. We also identified common biological pathways enriched in all three models, such as inflammation and immune response, which were also observed in tissue from patients. However, none of the models could recapitulate the profile of molecular changes observed in tissue from patients. SIGNIFICANCE: Our results indicate that each model has its own set of biological responses leading to epilepsy. Thus, it seems that only using a combination of the three models may one replicate more closely the mechanisms underlying MTLE+HS as seen in patients.
Subject(s)
Epilepsy, Temporal Lobe , Epilepsy , Animals , Benchmarking , Disease Models, Animal , Epilepsy/pathology , Epilepsy, Temporal Lobe/pathology , Humans , Proteome , Proteomics , SclerosisABSTRACT
BACKGROUND: Illumina's Infinium SNP BeadChips are extensively used in both small and large-scale genetic studies. A fundamental step in any analysis is the processing of raw allele A and allele B intensities from each SNP into genotype calls (AA, AB, BB). Various algorithms which make use of different statistical models are available for this task. We compare four methods (GenCall, Illuminus, GenoSNP and CRLMM) on data where the true genotypes are known in advance and data from a recently published genome-wide association study. RESULTS: In general, differences in accuracy are relatively small between the methods evaluated, although CRLMM and GenoSNP were found to consistently outperform GenCall. The performance of Illuminus is heavily dependent on sample size, with lower no call rates and improved accuracy as the number of samples available increases. For X chromosome SNPs, methods with sex-dependent models (Illuminus, CRLMM) perform better than methods which ignore gender information (GenCall, GenoSNP). We observe that CRLMM and GenoSNP are more accurate at calling SNPs with low minor allele frequency than GenCall or Illuminus. The sample quality metrics from each of the four methods were found to have a high level of agreement at flagging samples with unusual signal characteristics. CONCLUSIONS: CRLMM, GenoSNP and GenCall can be applied with confidence in studies of any size, as their performance was shown to be invariant to the number of samples available. Illuminus on the other hand requires a larger number of samples to achieve comparable levels of accuracy and its use in smaller studies (50 or fewer individuals) is not recommended.
Subject(s)
Algorithms , Genotype , Models, Statistical , Polymorphism, Single Nucleotide , Alleles , Cluster Analysis , Gene Frequency , Genome-Wide Association Study , Humans , Sample SizeABSTRACT
MOTIVATION: The availability of flexible open source software for the analysis of gene expression raw level data has greatly facilitated the development of widely used preprocessing methods for these technologies. However, the expansion of microarray applications has exposed the limitation of existing tools. RESULTS: We developed the oligo package to provide a more general solution that supports a wide range of applications. The package is based on the BioConductor principles of transparency, reproducibility and efficiency of development. It extends the existing tools and leverages existing code for visualization, accessing data and widely used preprocessing routines. The oligo package implements a unified paradigm for preprocessing data and interfaces with other BioConductor tools for downstream analysis. Our infrastructure is general and can be used by other BioConductor packages. AVAILABILITY: The oligo package is freely available through BioConductor, http://www.bioconductor.org.
Subject(s)
Gene Expression Profiling/methods , Oligonucleotide Array Sequence Analysis/methods , Software , Algorithms , Databases, Factual , Oligonucleotides , User-Computer InterfaceABSTRACT
MOTIVATION: Genome-wide association studies (GWAS) are used to discover genes underlying complex, heritable disorders for which less powerful study designs have failed in the past. The number of GWAS has skyrocketed recently with findings reported in top journals and the mainstream media. Microarrays are the genotype calling technology of choice in GWAS as they permit exploration of more than a million single nucleotide polymorphisms (SNPs) simultaneously. The starting point for the statistical analyses used by GWAS to determine association between loci and disease is making genotype calls (AA, AB or BB). However, the raw data, microarray probe intensities, are heavily processed before arriving at these calls. Various sophisticated statistical procedures have been proposed for transforming raw data into genotype calls. We find that variability in microarray output quality across different SNPs, different arrays and different sample batches have substantial influence on the accuracy of genotype calls made by existing algorithms. Failure to account for these sources of variability can adversely affect the quality of findings reported by the GWAS. RESULTS: We developed a method based on an enhanced version of the multi-level model used by CRLMM version 1. Two key differences are that we now account for variability across batches and improve the call-specific assessment of each call. The new model permits the development of quality metrics for SNPs, samples and batches of samples. Using three independent datasets, we demonstrate that the CRLMM version 2 outperforms CRLMM version 1 and the algorithm provided by Affymetrix, Birdseed. The main advantage of the new approach is that it enables the identification of low-quality SNPs, samples and batches. AVAILABILITY: Software implementing of the method described in this article is available as free and open source code in the crlmm R/BioConductor package. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
Subject(s)
Algorithms , Computational Biology/methods , Genotype , Genome-Wide Association Study , Oligonucleotide Array Sequence Analysis/methods , Polymorphism, Single NucleotideABSTRACT
UNLABELLED: Illumina produces a number of microarray-based technologies for human genotyping. An Infinium BeadChip is a two-color platform that types between 10(5) and 10(6) single nucleotide polymorphisms (SNPs) per sample. Despite being widely used, there is a shortage of open source software to process the raw intensities from this platform into genotype calls. To this end, we have developed the R/Bioconductor package crlmm for analyzing BeadChip data. After careful preprocessing, our software applies the CRLMM algorithm to produce genotype calls, confidence scores and other quality metrics at both the SNP and sample levels. We provide access to the raw summary-level intensity data, allowing users to develop their own methods for genotype calling or copy number analysis if they wish. AVAILABILITY AND IMPLEMENTATION: The crlmm Bioconductor package is available from http://www.bioconductor.org. Data packages and documentation are available from http://rafalab.jhsph.edu/software.html.
Subject(s)
Computational Biology/methods , Genome , Genotype , Software , AlgorithmsABSTRACT
The development of precision medicine strategies requires prior knowledge of the genetic background of the target population. However, despite the availability of data from admixed Americans within large reference population databases, we cannot use these data as a surrogate for that of the Brazilian population. This lack of transferability is mainly due to differences between ancestry proportions of Brazilian and other admixed American populations. To address the issue, a coalition of research centres created the Brazilian Initiative on Precision Medicine (BIPMed). In this study, we aim to characterise two datasets obtained from 358 individuals from the BIPMed using two different platforms: whole-exome sequencing (WES) and a single nucleotide polymorphism (SNP) array. We estimated allele frequencies and variant pathogenicity values from the two datasets and compared our results using the BIPMed dataset with other public databases. Here, we show that the BIPMed WES dataset contains variants not included in dbSNP, including 6480 variants that have alternative allele frequencies (AAFs) >1%. Furthermore, after merging BIPMed WES and SNP array data, we identified 809,589 variants (47.5%) not present within the 1000 Genomes dataset. Our results demonstrate that, through the incorporation of Brazilian individuals into public genomic databases, BIPMed not only was able to provide valuable knowledge needed for the implementation of precision medicine but may also enhance our understanding of human genome variability and the relationship between genetic variation and disease predisposition.
ABSTRACT
Antipsychotics are the main line of treatment for schizophrenia. Even though there are significant rates of medication drop out due to side effects and limited response of approximately 50% of patients. This is likely due to incomplete knowledge in how these drugs act at the molecular level. To improve treatment efficacy during the critical early stages of schizophrenia, we aimed to identify molecular signatures at baseline (T0) for prediction of a positive response to the atypical antipsychotics olanzapine and risperidone after 6 weeks (T6) treatment. Blood plasma samples were processed and analyzed by label-free quantitative shotgun proteomics using two-dimensional nano-liquid chromatography, coupled online to a Synapt G2-Si mass spectrometer. Data were obtained in MSE mode (data-independent acquisition) in combination with ion-mobility (HDMSE). We were able to identify a potential panel of proteins that might predict a positive outcome to olanzapine and risperidone treatment. The proteins found to be differentially abundant between T0 and T6 in good responders compared to poor responders were analyzed in silico for enrichment pathways and found to be mostly involved with immune system functions. This data can contribute to better understand the biochemical signaling mechanisms peripherally triggered by antipsychotic medication and eventually used to develop surrogate biomarker tests to help improve treatment outcomes and guide development of new treatment approaches. SIGNIFICANCE: The application of proteomics to the study of the atypical antipsychotic effects on the blood plasma proteome from schizophrenia patients could help in the search for new targets to improve the current therapies, as well as in the development of new therapeutic strategies. In this original article, we provided clues that atypical antipsychotics might be associated with good response by modulating proteins that play a role in inflammation and/or immune system pathways. In addition, the proteins with differential abundance found in the comparison between good and poor responders at the baseline might compose a signature for prediction of response effectiveness.
Subject(s)
Antipsychotic Agents , Schizophrenia , Antipsychotic Agents/therapeutic use , Benzodiazepines/therapeutic use , Humans , Olanzapine/therapeutic use , Plasma , Proteomics , Risperidone/therapeutic use , Schizophrenia/drug therapyABSTRACT
Mesial temporal lobe epilepsy (MTLE) is a chronic neurological disorder affecting almost 40% of adult patients with epilepsy. Hippocampal sclerosis (HS) is a common histopathological abnormality found in patients with MTLE. HS is characterised by extensive neuronal loss in different hippocampus sub-regions. In this study, we used laser microdissection-based microproteomics to determine the protein abundances in different regions and layers of the hippocampus dentate gyrus (DG) in an electric stimulation rodent model which displays classical HS damage similar to that found in patients with MTLE. Our results indicate that there are differences in the proteomic profiles of different layers (granule cell and molecular), as well as different regions, of the DG (ventral and dorsal). We have identified new signalling pathways and proteins present in specific layers and regions of the DG, such as PARK7, RACK1, and connexin 31/gap junction. We also found two major signalling pathways that are common to all layers and regions: inflammation and energy metabolism. Finally, our results highlight the utility of high-throughput microproteomics and spatial-limited isolation of tissues in the study of complex disorders to fully appreciate the large biological heterogeneity present in different cell populations within the central nervous system.
Subject(s)
Connexins/metabolism , Epilepsy, Temporal Lobe/metabolism , Hippocampus/metabolism , Protein Deglycase DJ-1/metabolism , Proteomics/methods , Receptors for Activated C Kinase/metabolism , Animals , Disease Models, Animal , Epilepsy, Temporal Lobe/etiology , Gene Expression Regulation , Humans , Laser Capture Microdissection , Organ Specificity , Protein Interaction Maps , Rats , Signal TransductionABSTRACT
Objective: We aimed to improve the classification of SCN1A missense variants in patients with Dravet syndrome (DS) by combining and modifying the current variants classification criteria to minimize inconclusive test results. Methods: We established a score classification workflow based on evidence of pathogenicity to adapt the classification of DS-related SCN1A missense variants. In addition, we compiled the variants reported in the literature and our cohort and assessed the proposed pathogenic classification criteria. We combined information regarding previously established pathogenic amino acid changes, mode of inheritance, population-specific allele frequencies, localization within protein domains, and deleterious effect prediction analysis. Results: Our meta-analysis showed that 46% (506/1,101) of DS-associated SCN1A variants are missense. We applied the score classification workflow and 56.5% (286/506) of the variants had their classification changed from VUS: 17.8% (90/506) into "pathogenic" and 38.7% (196/506) as "likely pathogenic." Conclusion: Our results indicate that using multimodal analysis seems to be the best approach to interpret the pathogenic impact of SCN1A missense changes for the molecular diagnosis of patients with DS. By applying the proposed workflow, most DS related SCN1A variants had their classification improved.
ABSTRACT
Admixed American populations have different global proportions of European, Sub-Saharan African, and Native-American ancestry. However, individuals who display the same global ancestry could exhibit remarkable differences in the distribution of local ancestry blocks. We studied for the first time the distribution of local ancestry across the genome of 264 Brazilian admixed individuals, ascertained within the scope of the Brazilian Initiative on Precision Medicine. We found a decreased proportion of European ancestry together with an excess of Native-American ancestry on chromosome 8p23.1 and showed that this is due to haplotypes created by chromosomal inversion events. Furthermore, Brazilian non-inverted haplotypes were more similar to Native-American haplotypes than to European haplotypes, in contrast to what was found in other American admixed populations. We also identified signals of recent positive selection on chromosome 8p23.1, and one gene within this locus, PPP1R3B, is related to glycogenesis and has been associated with an increased risk of type 2 diabetes and obesity. These findings point to a selection event after admixture, which is still not entirely understood in recent admixture events.
Subject(s)
Adaptation, Physiological/genetics , Black People/genetics , Brazil , Chromosomes/genetics , Diabetes Mellitus, Type 2/genetics , Female , Genetics, Population/methods , Genome, Human/genetics , Haplotypes/genetics , Humans , Male , Protein Phosphatase 1/genetics , United States , White People/geneticsABSTRACT
Mesial temporal lobe epilepsy is the most common form of adult epilepsy in surgical series. Currently, the only characteristic used to predict poor response to clinical treatment in this syndrome is the presence of hippocampal sclerosis. Single nucleotide polymorphisms (SNPs) located in genes encoding drug transporter and metabolism proteins could influence response to therapy. Therefore, we aimed to evaluate whether combining information from clinical variables as well as SNPs in candidate genes could improve the accuracy of predicting response to drug therapy in patients with mesial temporal lobe epilepsy. For this, we divided 237 patients into two groups: 75 responsive and 162 refractory to antiepileptic drug therapy. We genotyped 119 SNPs in ABCB1, ABCC2, CYP1A1, CYP1A2, CYP1B1, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4, and CYP3A5 genes. We used 98 additional SNPs to evaluate population stratification. We assessed a first scenario using only clinical variables and a second one including SNP information. The random forests algorithm combined with leave-one-out cross-validation was used to identify the best predictive model in each scenario and compared their accuracies using the area under the curve statistic. Additionally, we built a variable importance plot to present the set of most relevant predictors on the best model. The selected best model included the presence of hippocampal sclerosis and 56 SNPs. Furthermore, including SNPs in the model improved accuracy from 0.4568 to 0.8177. Our findings suggest that adding genetic information provided by SNPs, located on drug transport and metabolism genes, can improve the accuracy for predicting which patients with mesial temporal lobe epilepsy are likely to be refractory to drug treatment, making it possible to identify patients who may benefit from epilepsy surgery sooner.