| 1. |
- Romagnoni, A, et al.
(författare)
-
Comparative performances of machine learning methods for classifying Crohn Disease patients using genome-wide genotyping data
- 2019
-
Ingår i: Scientific reports. - : Springer Science and Business Media LLC. - 2045-2322. ; 9:1, s. 10351-
-
Tidskriftsartikel (refereegranskat)abstract
- Crohn Disease (CD) is a complex genetic disorder for which more than 140 genes have been identified using genome wide association studies (GWAS). However, the genetic architecture of the trait remains largely unknown. The recent development of machine learning (ML) approaches incited us to apply them to classify healthy and diseased people according to their genomic information. The Immunochip dataset containing 18,227 CD patients and 34,050 healthy controls enrolled and genotyped by the international Inflammatory Bowel Disease genetic consortium (IIBDGC) has been re-analyzed using a set of ML methods: penalized logistic regression (LR), gradient boosted trees (GBT) and artificial neural networks (NN). The main score used to compare the methods was the Area Under the ROC Curve (AUC) statistics. The impact of quality control (QC), imputing and coding methods on LR results showed that QC methods and imputation of missing genotypes may artificially increase the scores. At the opposite, neither the patient/control ratio nor marker preselection or coding strategies significantly affected the results. LR methods, including Lasso, Ridge and ElasticNet provided similar results with a maximum AUC of 0.80. GBT methods like XGBoost, LightGBM and CatBoost, together with dense NN with one or more hidden layers, provided similar AUC values, suggesting limited epistatic effects in the genetic architecture of the trait. ML methods detected near all the genetic variants previously identified by GWAS among the best predictors plus additional predictors with lower effects. The robustness and complementarity of the different methods are also studied. Compared to LR, non-linear models such as GBT or NN may provide robust complementary approaches to identify and classify genetic markers.
|
|
| 2. |
- Momozawa, Y, et al.
(författare)
-
IBD risk loci are enriched in multigenic regulatory modules encompassing putative causative genes
- 2018
-
Ingår i: Nature communications. - : Springer Science and Business Media LLC. - 2041-1723. ; 9:1, s. 2427-
-
Tidskriftsartikel (refereegranskat)abstract
- GWAS have identified >200 risk loci for Inflammatory Bowel Disease (IBD). The majority of disease associations are known to be driven by regulatory variants. To identify the putative causative genes that are perturbed by these variants, we generate a large transcriptome data set (nine disease-relevant cell types) and identify 23,650 cis-eQTL. We show that these are determined by ∼9720 regulatory modules, of which ∼3000 operate in multiple tissues and ∼970 on multiple genes. We identify regulatory modules that drive the disease association for 63 of the 200 risk loci, and show that these are enriched in multigenic modules. Based on these analyses, we resequence 45 of the corresponding 100 candidate genes in 6600 Crohn disease (CD) cases and 5500 controls, and show with burden tests that they include likely causative genes. Our analyses indicate that ≥10-fold larger sample sizes will be required to demonstrate the causality of individual genes using this approach.
|
|
| 3. |
|
|
| 4. |
- Cleynen, Isabelle, et al.
(författare)
-
Inherited determinants of Crohn's disease and ulcerative colitis phenotypes : a genetic association study
- 2016
-
Ingår i: The Lancet. - New York, USA : Elsevier. - 0140-6736 .- 1474-547X. ; 387:10014, s. 156-167
-
Tidskriftsartikel (refereegranskat)abstract
- Background: Crohn's disease and ulcerative colitis are the two major forms of inflammatory bowel disease; treatment strategies have historically been determined by this binary categorisation. Genetic studies have identified 163 susceptibility loci for inflammatory bowel disease, mostly shared between Crohn's disease and ulcerative colitis. We undertook the largest genotype association study, to date, in widely used clinical subphenotypes of inflammatory bowel disease with the goal of further understanding the biological relations between diseases.Methods This study included patients from 49 centres in 16 countries in Europe, North America, and Australasia. We applied the Montreal classification system of inflammatory bowel disease subphenotypes to 34,819 patients (19,713 with Crohn's disease, 14,683 with ulcerative colitis) genotyped on the Immunochip array. We tested for genotype-phenotype associations across 156,154 genetic variants. We generated genetic risk scores by combining information from all known inflammatory bowel disease associations to summarise the total load of genetic risk for a particular phenotype. We used these risk scores to test the hypothesis that colonic Crohn's disease, ileal Crohn's disease, and ulcerative colitis are all genetically distinct from each other, and to attempt to identify patients with a mismatch between clinical diagnosis and genetic risk profile.Findings: After quality control, the primary analysis included 29,838 patients (16,902 with Crohn's disease, 12,597 with ulcerative colitis). Three loci (NOD2, MHC, and MST1 3p21) were associated with subphenotypes of inflammatory bowel disease, mainly disease location (essentially fixed over time; median follow-up of 10·5 years). Little or no genetic association with disease behaviour (which changed dramatically over time) remained after conditioning on disease location and age at onset. The genetic risk score representing all known risk alleles for inflammatory bowel disease showed strong association with disease subphenotype (p=1·65 × 10(-78)), even after exclusion of NOD2, MHC, and 3p21 (p=9·23 × 10(-18)). Predictive models based on the genetic risk score strongly distinguished colonic from ileal Crohn's disease. Our genetic risk score could also identify a small number of patients with discrepant genetic risk profiles who were significantly more likely to have a revised diagnosis after follow-up (p=6·8 × 10(-4)).Interpretation: Our data support a continuum of disorders within inflammatory bowel disease, much better explained by three groups (ileal Crohn's disease, colonic Crohn's disease, and ulcerative colitis) than by Crohn's disease and ulcerative colitis as currently defined. Disease location is an intrinsic aspect of a patient's disease, in part genetically determined, and the major driver to changes in disease behaviour over time.Funding: International Inflammatory Bowel Disease Genetics Consortium members funding sources (see Acknowledgments for full list).
|
|
| 5. |
- Turkington, RC, et al.
(författare)
-
Immune activation by DNA damage predicts response to chemotherapy and survival in oesophageal adenocarcinoma
- 2019
-
Ingår i: Gut. - : BMJ. - 1468-3288 .- 0017-5749. ; 68:11, s. 1918-1927
-
Tidskriftsartikel (refereegranskat)abstract
- Current strategies to guide selection of neoadjuvant therapy in oesophageal adenocarcinoma (OAC) are inadequate. We assessed the ability of a DNA damage immune response (DDIR) assay to predict response following neoadjuvant chemotherapy in OAC.DesignTranscriptional profiling of 273 formalin-fixed paraffin-embedded prechemotherapy endoscopic OAC biopsies was performed. All patients were treated with platinum-based neoadjuvant chemotherapy and resection between 2003 and 2014 at four centres in the Oesophageal Cancer Clinical and Molecular Stratification consortium. CD8 and programmed death ligand 1 (PD-L1) immunohistochemical staining was assessed in matched resection specimens from 126 cases. Kaplan-Meier and Cox proportional hazards regression analysis were applied according to DDIR status for recurrence-free survival (RFS) and overall survival (OS).ResultsA total of 66 OAC samples (24%) were DDIR positive with the remaining 207 samples (76%) being DDIR negative. DDIR assay positivity was associated with improved RFS (HR: 0.61; 95% CI 0.38 to 0.98; p=0.042) and OS (HR: 0.52; 95% CI 0.31 to 0.88; p=0.015) following multivariate analysis. DDIR-positive patients had a higher pathological response rate (p=0.033), lower nodal burden (p=0.026) and reduced circumferential margin involvement (p=0.007). No difference in OS was observed according to DDIR status in an independent surgery-alone dataset.DDIR-positive OAC tumours were also associated with the presence of CD8+ lymphocytes (intratumoural: p<0.001; stromal: p=0.026) as well as PD-L1 expression (intratumoural: p=0.047; stromal: p=0.025).ConclusionThe DDIR assay is strongly predictive of benefit from DNA-damaging neoadjuvant chemotherapy followed by surgical resection and is associated with a proinflammatory microenvironment in OAC.
|
|
| 6. |
- Ellinghaus, David, et al.
(författare)
-
Analysis of five chronic inflammatory diseases identifies 27 new associations and highlights disease-specific patterns at shared loci
- 2016
-
Ingår i: Nature Genetics. - New York, USA : Nature Publishing Group. - 1061-4036 .- 1546-1718. ; 48:5, s. 510-518
-
Tidskriftsartikel (refereegranskat)abstract
- We simultaneously investigated the genetic landscape of ankylosing spondylitis, Crohn's disease, psoriasis, primary sclerosing cholangitis and ulcerative colitis to investigate pleiotropy and the relationship between these clinically related diseases. Using high-density genotype data from more than 86,000 individuals of European ancestry, we identified 244 independent multidisease signals, including 27 new genome-wide significant susceptibility loci and 3 unreported shared risk loci. Complex pleiotropy was supported when contrasting multidisease signals with expression data sets from human, rat and mouse together with epigenetic and expressed enhancer profiles. The comorbidities among the five immune diseases were best explained by biological pleiotropy rather than heterogeneity (a subgroup of cases genetically identical to those with another disease, possibly owing to diagnostic misclassification, molecular subtypes or excessive comorbidity). In particular, the strong comorbidity between primary sclerosing cholangitis and inflammatory bowel disease is likely the result of a unique disease, which is genetically distinct from classical inflammatory bowel disease phenotypes.
|
|
| 7. |
- Colver, A., et al.
(författare)
-
Self-reported quality of life of adolescents with cerebral palsy: a cross-sectional and longitudinal analysis
- 2015
-
Ingår i: Lancet. - : Elsevier BV. - 0140-6736. ; 385:9969, s. 705-716
-
Tidskriftsartikel (refereegranskat)abstract
- Background Children with cerebral palsy who can self-report have similar quality of life (QoL) to their able-bodied peers. Is this similarity also found in adolescence? We examined how self-reported QoL of adolescents with cerebral palsy varies with impairment and compares with the general population, and how factors in childhood predict adolescent QoL. Methods We report QoL outcomes in a longitudinal follow-up and cross-sectional analysis of individuals included in the SPARCLE1 (childhood) and SPARCLE2 (adolescent) studies. In 2004 (SPARCLE1), a cohort of 818 children aged 8-12 years were randomly selected from population-based cerebral palsy registers in nine European regions. We gathered data from 500 participants about QoL with KIDSCREEN (ten domains); frequency of pain; child psychological problems (Strengths and Difficulties Questionnaire); and parenting stress (Parenting Stress Index). At follow-up in 2009 (SPARCLE2), 355 (71%) adolescents aged 13-17 years remained in the study and self-reported QoL (longitudinal sample). 76 additional adolescents self-reported QoL in 2009, providing data for 431 adolescents in the cross-sectional sample. Researchers gathered data at home visits. We compared QoL against matched controls in the general population. We used multivariable regression to relate QoL of adolescents with cerebral palsy to impairments (cross-sectional analysis) and to childhood QoL, pain, psychological problems, and parenting stress (longitudinal analysis). Findings Severity of impairment was significantly associated (p<0.01) with reduced adolescent QoL on only three domains (Moods and emotions, Autonomy, and Social support and peers); average differences in QoL between the least and most able groups were generally less than 0.5 SD. Adolescents with cerebral palsy had significantly lower QoL than did those in the general population in only one domain (Social support and peers; mean difference -2.7 [0.25 SD], 95% CI -4.3 to -1.4). Pain in childhood or adolescence was strongly associated with low adolescent QoL on eight domains. Childhood QoL was a consistent predictor of adolescent QoL. Child psychological problems and parenting stress in childhood or their worsening between childhood and adolescence predicted only small reductions in adolescent QoL. Interpretation Individual and societal attitudes should be affected by the similarity of the QoL of adolescents with and without cerebral palsy. Adolescents with cerebral palsy need particular help to maintain and develop peer relationships. Interventions in childhood to alleviate psychological difficulties, parenting stress, and especially pain, are justified for their intrinsic value and for their longer term effect on adolescent QoL. Copyright (C) Colver et al. Open Access article distributed under the terms of CC BY.
|
|
| 8. |
- Dang, V. M., et al.
(författare)
-
Predictors of participation of adolescents with cerebral palsy: A European multi-centre longitudinal study
- 2015
-
Ingår i: Research in Developmental Disabilities. - : Elsevier BV. - 0891-4222. ; 36, s. 551-564
-
Tidskriftsartikel (refereegranskat)abstract
- We investigated whether childhood factors that are amenable to intervention (parenting stress, child psychological problems and pain) predicted participation in daily activities and social roles of adolescents with cerebral palsy (CP). We randomly selected 1174 children aged 8-12 years from eight population-based registers of children with CP in six European countries; 743 (63%) agreed to participate. One further region recruited 75 children from multiple sources. These 818 children were visited at home at age 8-12 years, 594 (73%) agreed to follow-up at age 13-17 years. We used the following measures: parent reported stress (Parenting Stress Index Short Form), their child's psychological difficulties (Strength and Difficulties Questionnaire) and frequency and severity of pain; either child or parent reported the child's participation (LIFE Habits questionnaire). We fitted a structural equation model to each of the participation domains, regressing participation in childhood and adolescence on parenting stress, child psychological problems and pain, and regressing adolescent factors on the corresponding childhood factors; models were adjusted for impairment, region, age and gender. Pain in childhood predicted restricted adolescent participation in all domains except Mealtimes and Communication (standardized total indirect effects beta -0.05 to -0.18, 0.01 < p < 0.05 to p < 0.001, depending on domain). Psychological problems in childhood predicted restricted adolescent participation in all domains of social roles, and in Personal Care and Communication (beta -0.07 to -0.17,0.001 < p < 0.01 top < 0.001). Parenting stress in childhood predicted restricted adolescent participation in Health Hygiene, Mobility and Relationships (beta -0.07 to -0.18, 0.001 < p < 0.01 to p < 0.001). These childhood factors predicted adolescent participation largely via their effects on childhood participation; though in some domains early psychological problems and parenting stress in childhood predicted adolescent participation largely through their persistence into adolescence. We conclude that participation of adolescents with CP was predicted by early modifiable factors related to the child and family. Interventions for reduction of pain, psychological difficulties and parenting stress in childhood are justified not only for their intrinsic value, but also for probable benefits to childhood and adolescent participation. (C) 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/3.0/).
|
|
| 9. |
- Dimitrakakis, Christos, 1975, et al.
(författare)
-
Bayesian Fairness
- 2019
-
Ingår i: THIRTY-THIRD AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE / THIRTY-FIRST INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE / NINTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE. - 9781577358091 ; , s. 509-516
-
Konferensbidrag (refereegranskat)abstract
- We consider the problem of how decision making can be fair when the underlying probabilistic model of the world is not known with certainty. We argue that recent notions of fairness in machine learning need to explicitly incorporate parameter uncertainty, hence we introduce the notion of Bayesian fairness as a suitable candidate for fair decision rules. Using balance, a definition of fairness introduced in (Kleinberg, Mullainathan, and Raghavan 2016), we show how a Bayesian perspective can lead to well-performing and fair decision rules even under high uncertainty.
|
|
| 10. |
- Dimitrakakis, Christos, 1975, et al.
(författare)
-
Multi-view decision processes: The helper-AI problem
- 2017
-
Ingår i: Advances in Neural Information Processing Systems. - 1049-5258. ; 2017-December, s. 5444-5453
-
Konferensbidrag (refereegranskat)abstract
- © 2017 Neural information processing systems foundation. All rights reserved. We consider a two-player sequential game in which agents have the same reward function but may disagree on the transition probabilities of an underlying Markovian model of the world. By committing to play a specific policy, the agent with the correct model can steer the behavior of the other agent, and seek to improve utility. We model this setting as a multi-view decision process, which we use to formally analyze the positive effect of steering policies. Furthermore, we develop an algorithm for computing the agents' achievable joint policy, and we experimentally show that it can lead to a large utility increase when the agents' models diverge.
|
|
