Search

Table S1: Data set country allocation; Table S2. A Cox regression model for 2 year overall survival considering top predictors- no interactions considered; Table S3. A Cox regression model for 2 year overall survival considering top predictors with interactions considered; Table S4: Population characteristics of a randomly selected imputed data set; Table S5. Risk for 2-years OS and 2-years LFS categorized by the UCBT score; Table S6. Probabilities for OS, LFS, NRM, and RI stratified by the UCBT score over 20 imputed datasets; Table S7: The population characteristics across the UCBT risk score categories over the validation set; Figure S1. Estimation of predictor importance using the Random Survival Forest minimal depth; Figure S2: Interaction analysis using conditional plots; Figure S3: A calibration plot of the risk scores between predicted and observed events (death) at two years, according to pentiles.

Purpose: Survival of acute leukemia (AL) patients following umbilical cord blood transplantation (UCBT) is dependent on an array of individual features. Integrative models for risk assessment are lacking. We sought to develop a scoring system for prediction of overall survival (OS) and leukemia-free survival (LFS) at 2 years following UCBT in AL patients.Experimental Design: The study cohort included 3,140 pediatric and adult AL UCBT patients from the European Society of Blood and Marrow Transplantation and Eurocord registries. Patients received single or double cord blood units. The dataset was geographically split into a derivation (n = 2,362, 65%) and validation set (n = 778, 35%). Top predictors of OS were identified using the Random Survival Forest algorithm and introduced into a Cox regression model, which served for the construction of the UCBT risk score.Results: The score includes nine variables: disease status, diagnosis, cell dose, age, center experience, cytomegalovirus serostatus, degree of HLA mismatch, previous autograft, and anti-thymocyte globulin administration. Over the validation set an increasing score was associated with decreasing probabilities for 2 years OS and LFS, ranging from 70.21% [68.89–70.71, 95% confidence interval (CI)] and 64.76% (64.33–65.86, 95% CI) to 14.78% (10.91–17.41) and 18.11% (14.40–22.30), respectively. It stratified patients into six distinct risk groups. The score's discrimination (AUC) over multiple imputations of the validation set was 68.76 (68.19–69.04, range) and 65.78 (65.20–66.28) for 2 years OS and LFS, respectively.Conclusions: The UCBT score is a simple tool for risk stratification of AL patients undergoing UCBT. Widespread application of the score will require further independent validation. Clin Cancer Res; 23(21); 6478–86. ©2017 AACR.

Recently, in parallel to decrease in semen quality, the consumption of sugar has risen sharply. This provided the rationale to study the association between whole dietary sugar consumption and semen quality. Our aim was to investigate the association between sugar consumption and semen quality. The final cross-sectional study population (n = 280 of initial n = 593, after applying inclusion and exclusion criteria) attending routine semen analysis at sperm bank laboratory was subject to semen quality analysis according to WHO criteria (volume, sperm concentration, total sperm count, percentage total motility, and percentage normal morphology) and filled food frequency questionnaire (FFQ) and lifestyle questionnaire. Associations between consumed sugars and semen quality were analyzed using multivariate regression adjusted to relevant cofounders for 2 food components containing sugar including soft drinks (SoftD) and total added sugar to food products (SugProd). We found negative association between higher consumption of dietary sugar in all 2 dietary sub-categories and sperm concentration. Significant sperm concentration decrements of 18% and 23% were associated with SoftD median consumption of 0.2 drinks/day (IQR; 0.1-0.5 drinks/day). Significant sperm concentration decrements of 15% and 17% were associated with median SugProd consumption of 25 teaspoons of added sugar/day (IQR; 19-31 teaspoons of added sugar/day). In conclusion, our study findings demonstrate that sugar consumption is negatively associated with sperm concentration.

Background Various prognostic models for mortality prediction following ST-segment elevation myocardial infarction (STEMI) have been developed over the past two decades. Our group has previously demonstrated that machine learning (ML)-based models can outperform known risk scores for 30-day mortality post-STEMI. The study aimed to redevelop an ML-based random forest prediction model for 30-day mortality post-STEMI and externally validate it on a large cohort. Methods This was a retrospective, supervised learning, data mining study developed on the Acute Coronary Syndrome Israeli Survey (ACSIS) registry and the Myocardial Ischemia National Audit Project (MINAP) for external validation. Patients included received reperfusion therapy for STEMI between 2006 and 2016. Discrimination and calibration performances were assessed for two developed models and compared with the Global Registry of Acute Cardiac Events (GRACE) score. Results The ACSIS cohort (2,782 included /15,212 total) and MINAP cohort (22,693 included/735,000 total) were significantly different in most variables, yet similar in 30-day mortality rate (4.3–4.4%). Random forest models were developed on the ACSIS cohort with a full model including all 32 variables and a simple model including the 10 most important ones. Features’ importance was calculated using the varImp function measuring how much each feature contributes to the data's homogeneity. Applying the optimized models on the MINAP validation cohort showed high discrimination of area under the curve (AUC) = 0.804 (0.786–0.822) for the full model, and AUC = 0.787 (0.748–0.780) using the simple model, compared with the GRACE risk score discrimination of AUC = 0.764 (0.748–0.780). All models were not well calibrated for the MINAP data. Following Platt scaling on 20% of the MINAP data, the random forest models calibration improved while the GRACE calibration did not change. Conclusions The random forest predictive model for 30-day mortality post STEMI, developed on the ACSIS national registry, has been validated in the MINAP large external cohort and can be applied early at admission for risk stratification. The model performed better than the commonly used GRACE score. Furthermore, to the best of our knowledge, this is the first externally validated ML-based model for STEMI.

INTRODUCTION: Molecular mechanisms underlying the well-known association between Type 2 diabetes (T2D) and late-life cognitive impairment are unknown. Alterations in glyco post-translational modifications (PTM) of serum proteins reflect altered cellular physiology and are associated with T2D and impaired cognition. This novel large-scale mass spectrometry based glycoproteomics study aimed to identify therapeutic PTM targets that may link T2D with poor cognition. METHODS: We performed LC-MS/MS–based novel glycoproteomic profiling of sera samples from 23 cognitively normal older adults with T2D. At baseline all people had normal cognition, but during 36-months, 8 participants developed cognitive impairment while 15 maintained normal cognition RESULTS: We identified and quantified a total of 1,866 unique glycoproteoforms (i.e., peptides with a post-transcription sugar modification) that matched 201 annotated proteins. Group differences were observed for 72 glycoproteoforms at baseline. At the time of cognitive testing three years later, there were group differences for 137 glycoproteoforms. Group differences were observed for 14 glycoproteoforms both at baseline and at 3-year follow-up. DISCUSSION: Glycoproteoforms may contribute to cognitive impairment in T2D, may identify adults with T2D at risk for cognitive impairment and are high value therapeutic targets for new therapies to mitigate the negative cognitive effects of T2D in old age.

SummaryRecently, Deep Learning models, initially developed in the field of Natural Language Processing (NLP), were applied successfully to analyze protein sequences. A major drawback of these models is their size in terms of the number of parameters needed to be fitted and the amount of computational resources they require. Recently, “distilled” models using the concept of student and teacher networks have been widely used in NLP. Here, we adapted this concept to the problem of protein sequence analysis, by developing DistilProtBert, a distilled version of the successful ProtBert model. Implementing this approach, we reduced the size of the network and the running time by 50%, and the computational resources needed for pretraining by 98% relative to ProtBert model. Using two published tasks, we showed that the performance of the distilled model approaches that of the full model. We next tested the ability of DistilProtBert to distinguish between real and random protein sequences. The task is highly challenging if the composition is maintained on the level of singlet, doublet and triplet amino acids. Indeed, traditional machine learning algorithms have difficulties with this task. Here, we show that DistilProtBert preforms very well on singlet, doublet, and even triplet-shuffled versions of the human proteome, with AUC of 0.92, 0.91, and 0.87 respectively. Finally, we suggest that by examining the small number of false-positive classifications (i.e., shuffled sequences classified as proteins by DistilProtBert) we may be able to identify de-novo potential natural-like proteins based on random shuffling of amino acid sequences.Availabilityhttps://github.com/yarongef/DistilProtBertContactyaron.geffen@biu.ac.il

Background High-grade gliomas (HGG) in children have a devastating prognosis and occur in a remarkable spatiotemporal pattern. Diffuse midline gliomas (DMG), including diffuse intrinsic pontine gliomas (DIPG), typically occur in mid-childhood, while cortical HGGs are more frequent in older children and adults. The mechanisms behind this pattern are not clear. Methods We used mouse organotypic slice cultures and glial cell cultures to test the impact of the microenvironment on human DIPG cells. Comparing the expression between brainstem and cortical microglia identified differentially expressed secreted proteins. The impact of some of these proteins on DIPGs was tested. Results DIPGs, pediatric HGGs of brainstem origin, survive and divide more in organotypic slice cultures originating in the brainstem as compared to the cortex. Moreover, brainstem microglia are better able to support tumors of brainstem origin. A comparison between the two microglial populations revealed differentially expressed genes. One such gene, interleukin-33 (IL33), is highly expressed in the pons of young mice and its DIPG receptor is upregulated in this context. Consistent with this observation, the expression levels of IL33 and its receptor, IL1RL1, are higher in DIPG biopsies compared to low-grade cortical gliomas. Furthermore, IL33 can enhance proliferation and clonability of HGGs of brainstem origin, while blocking IL33 in brainstem organotypic slice cultures reduced the proliferation of these tumor cells. Conclusions Crosstalk between DIPGs and the brainstem microenvironment, in particular microglia, through IL33 and other secreted factors, modulates spatiotemporal patterning of this HGG and could prove to be an important future therapeutic target.

Objective: Neonatal intensive care unit (NICU) admission among term neonates is associated with significant morbidity and mortality, as well as high healthcare costs. A comprehensive NICU admission risk assessment using an integrated statistical approach for this rare admission event may be used to build a risk calculation algorithm for this group of neonates prior to delivery. Methods: A single-center case–control retrospective study was conducted between August 2005 and December 2019, including in-hospital singleton live born neonates, born at ≥37 weeks’ gestation. Analyses included univariate and multivariable models combined with the machine learning gradient-boosting model (GBM). The primary aim of the study was to identify and quantify risk factors and causes of NICU admission of term neonates. Results: During the study period, 206,509 births were registered at the Shaare Zedek Medical Center. After applying the study exclusion criteria, 192,527 term neonates were included in the study; 5292 (2.75%) were admitted to the NICU. The NICU admission risk was significantly higher (ORs [95%CIs]) for offspring of nulliparous women (1.19 [1.07, 1.33]), those with diabetes mellitus or hypertensive complications of pregnancy (2.52 [2.09, 3.03] and 1.28 [1.02, 1.60] respectively), and for those born during the 37th week of gestation (2.99 [2.63, 3.41]; p < 0.001 for all), adjusted for congenital malformations and genetic syndromes. A GBM to predict NICU admission applied to data prior to delivery showed an area under the receiver operating characteristic curve of 0.750 (95%CI 0.743–0.757) and classified 27% as high risk and 73% as low risk. This risk stratification was significantly associated with adverse maternal and neonatal outcomes. Conclusion: The present study identified NICU admission risk factors for term neonates; along with the machine learning ranking of the risk factors, the highly predictive model may serve as a basis for individual risk calculation algorithm prior to delivery. We suggest that in the future, this type of planning of the delivery will serve different health systems, in both high- and low-resource environments, along with the NICU admission or transfer policy.

Objective To create a personalised machine learning model for prediction of severe adverse neonatal outcomes (SANO) during the second stage of labour. Design Retrospective Electronic-Medical-Record (EMR) -based study. Population A cohort of 73 868 singleton, term deliveries that reached the second stage of labour, including 1346 (1.8%) deliveries with SANO. Methods A gradient boosting model was created, analysing 21 million data points from antepartum features (e.g. gravidity and parity) gathered at admission to the delivery unit, and intrapartum data (e.g. cervical dilatation and effacement) gathered during the first stage of labour. Deliveries were allocated to high-risk and low-risk groups based on the Youden index to maximise sensitivity and specificity. Main outcome measures SANO was defined as either umbilical cord pH levels ≤7.1 or 1-minute or 5-minute Apgar score ≤7. Results The model for prediction of SANO yielded an area under the receiver operating curve (AUC) of 0.761 (95% CI 0.748-0.774). A third of the cohort (33.5%, n = 24 721) were allocated to a high-risk group for SANO, which captured up to 72.1% of these cases (odds ratio 5.3, 95% CI 4.7-6.0; high-risk versus low-risk groups). Conclusions Data acquired throughout the first stage of labour can be used to predict SANO during the second stage of labour using a machine learning model. Stratifying parturients at the beginning of the second stage of labour in a 'time out' session, can direct a personalised approach to management of this challenging aspect of labour, as well as improve allocation of staff and resources. Tweetable abstract Personalised prediction score for severe adverse neonatal outcomes in labour using machine learning model.

Obstetric anal sphincter injury (OASI) is a complication with substantial maternal morbidity. The aim of this study was to develop a machine learning model that would allow a personalized prediction algorithm for OASI, based on maternal and fetal variables collected at admission to labor. We performed a retrospective cohort study at a tertiary university hospital. Included were term deliveries (live, singleton, vertex). A comparison was made between women diagnosed with OASI and those without such injury. For formation of a machine learning-based model, a gradient boosting machine learning algorithm was implemented. Evaluation of the performance model was achieved using the area under the receiver-operating characteristic curve (AUC). Our cohort comprised 98,463 deliveries, of which 323 (0.3%) were diagnosed with OASI. Applying a machine learning model to data recorded during admission to labor allowed for individualized risk assessment with an AUC of 0.756 (95% CI 0.732–0.780). According to this model, a lower number of previous births, fewer pregnancies, decreased maternal weight and advanced gestational week elevated the risk for OASI. With regard to parity, women with one previous delivery had approximately 1/3 of the risk for OASI compared to nulliparous women (OR = 0.3 (0.23–0.39), p

The parasite Trypanosoma brucei cycles between insect and mammalian hosts, and is the causative agent of sleeping sickness. Here, we performed genome-wide mapping of 2ʹ-O-methylations (Nms) on trypanosome rRNA using three high-throughput sequencing methods; RibOxi-seq, RiboMeth-seq and 2ʹ-OMe-seq. This is the first study using three genome-wide mapping approaches on rRNA from the same species showing the discrepancy among the methods. RibOxi-seq detects all the sites, but RiboMeth-seq is the only method to evaluate the level of a single Nm site. The sequencing revealed at least ninety-nine Nms guided by eighty-five snoRNAs among these thirty-eight Nms are trypanosome specific sites. We present the sequence and target of the C/D snoRNAs guiding on rRNA. This is the highest number of Nms detected to date on rRNA of a single cell parasite. Based on RiboMeth-seq, several Nm sites were found to be differentially regulated at the two stages of the parasite life cycle, the insect procyclic form (PCF) versus the bloodstream form (BSF) in the mammalian host.

How genome instability is harnessed for fitness gain despite its potential deleterious effects is largely elusive. An ideal system to address this important open question is provided by the protozoan pathogen Leishmania, which exploits frequent variations in chromosome and gene copy number to regulate expression levels. Using ecological genomics and experimental evolution approaches we provide first evidence that Leishmania adaptation relies on epistatic interactions between functionally associated gene copy number variations in pathways driving fitness gain in a given environment. We further uncover post-transcriptional regulation as a key mechanism that compensates for deleterious gene dosage effects and provides phenotypic robustness to genetically heterogenous parasite populations. Finally, we correlate dynamic variations in snoRNA gene dosage with changes in rRNA 2’-O-methylation and pseudouridylation, suggesting translational control is an additional layer of parasite adaptation. Leishmania genome instability is thus harnessed for fitness gain by genome-dependent variations in gene expression, and genome-independent, compensatory mechanisms. This allows for polyclonal adaptation and maintenance of genetic heterogeneity despite strong selective pressure. The epistatic adaptation described here needs to be considered in Leishmania epidemiology and biomarker discovery, and may be relevant to other fast evolving, eukaryotic cells that exploit genome instability for adaptation, such as fungal pathogens or cancer.One Sentence SummaryEpistatic interactions harness genome instability for Leishmania fitness gain.

BACKGROUND Clinical risk assessment with quantitative formal risk scores may add to intuitive physician risk assessment and are advised by the international guidelines for the management of acute coronary syndrome (ACS) patients. Most previous studies have used the binary regression/classification approach (dead/alive) for long-term mortality post-ACS, without considering the time-to-event as in survival analysis. The use of machine learning (ML)-based survival models has yet to be validated. The primary objective was to compare survival prediction performance of 1-year mortality following ACS of two newly developed ML-based models [random survival forest (RSF) and deep learning (DeepSurv)] with the traditional Cox-proportional hazard (CPH) model. The secondary objective was external validation of the findings. METHODS This was a retrospective, supervised learning data mining study based on the Acute Coronary Syndrome Israeli Survey (ACSIS) and the Myocardial Ischemia National Audit Project (MINAP). The ACSIS data were divided to train/test in a 70/30 fashion. Next, the models were externally validated on the MINAP data. Harrell's C-index, inverse probability of censoring weighting (IPCW), and the Brier-score were used for models' performance comparison. RESULTS RSF performed best among the three models, with Harrell's C-index on training and testing sets reaching 0.953 and 0.924 respectively, followed by CPH multivariate selected model (0.805/0.849), CPH Univariate selected model (0.828/0.806), DeepSurv model (0.801/0.804), and the traditional CPH model (0.826/0.738). The RSF model also had the highest performance on the validation data set with 0.811 for Harrell's C-index, 0.844 for IPCW, and 0.093 for Brier score. The CPH model performance on the validation set had C-index range between 0.689 to 0.790, 0.713 to 0.826 for IPCW, and 0.094 to 0.103 Brier score. CONCLUSIONS RSF survival predictions for long-term mortality post-ACS show improved model performance compared with the classic statistical method. This may benefit patients by allowing better risk stratification and tailored therapy, however further prospective evaluations are required.

Precision medicine and sequence-based clinical diagnostics seek to predict disease risk or to identify causative variants from sequencing data. The Critical Assessment of Genome Interpretation (CAGI) is a community experiment consisting of genotype-phenotype prediction challenges; participants build models, undergo assessment, and share key findings. In the past, few CAGI challenges have addressed the impact of sequence variants on splicing. In CAGI5, two challenges (Vex-seq and MaPSY) involved prediction of the effect of variants, primarily single-nucleotide changes, on splicing. Although there are significant differences between these two challenges, both involved prediction of results from high-throughput exon inclusion assays. Here, we discuss the methods used to predict the impact of these variants on splicing, their performance, strengths, and weaknesses, and prospects for predicting the impact of sequence variation on splicing and disease phenotypes.

In trypanosomes, in contrast to most eukaryotes, the large subunit (LSU) ribosomal RNA is fragmented into two large and four small ribosomal RNAs (srRNAs) pieces, and this additional processing likely requires trypanosome-specific factors. Here, we examined the role of 10 abundant small nucleolar RNAs (snoRNAs) involved in rRNA processing. We show that each snoRNA involved in LSU processing associates with factors engaged in either early or late biogenesis steps. Five of these snoRNAs interact with the intervening sequences of rRNA precursor, whereas the others only guide rRNA modifications. The function of the snoRNAs was explored by silencing snoRNAs. The data suggest that the LSU rRNA processing events do not correspond to the order of rRNA transcription, and that srRNAs 2, 4 and 6 which are part of LSU are processed before srRNA1. Interestingly, the 6 snoRNAs that affect srRNA1 processing guide modifications on rRNA positions that span locations from the protein exit tunnel to the srRNA1, suggesting that these modifications may serve as check-points preceding the liberation of srRNA1. This study identifies the highest number of snoRNAs so far described that are involved in rRNA processing and/or rRNA folding and highlights their function in the unique trypanosome rRNA maturation events.

The GroE chaperonin system facilitates protein folding in an ATP-dependent manner. It has remained unclear why some proteins are obligate clients of the GroE system, whereas other closely related proteins are able to fold efficiently in its absence. Factors that cause folding to be slower affect kinetic partitioning between spontaneous folding and chaperone binding in favor of the latter. One such potential factor is contact order (CO), which is the average separation in sequence between residues that are in contact in the native structure. Here, we generated variants of enhanced green fluorescent protein with different COs using circular permutations. We found that GroE dependence in vitro and in vivo increases with increasing CO. Thus, our results show that CO is relevant not only for folding in vitro of relatively simple model systems but also for chaperonin dependence and folding in vivo.

Trypanosoma brucei, the parasite that causes sleeping sickness, cycles between an insect and a mammalian host. However, the effect of RNA modifications such as pseudouridinylation on its ability to survive in these two different host environments is unclear. Here, two genome-wide approaches were applied for mapping pseudouridinylation sites (Ψs) on small nucleolar RNA (snoRNA), 7SL RNA, vault RNA, and tRNAs from T. brucei. We show using HydraPsiSeq and RiboMeth-seq that the Ψ on C/D snoRNA guiding 2'-O-methylation increased the efficiency of the guided modification on its target, rRNA. We found differential levels of Ψs on these noncoding RNAs in the two life stages (insect host and mammalian host) of the parasite. Furthermore, tRNA isoform abundance and Ψ modifications were characterized in these two life stages demonstrating stage-specific regulation. We conclude that the differential Ψ modifications identified here may contribute to modulating the function of noncoding RNAs involved in rRNA processing, rRNA modification, protein synthesis, and protein translocation during cycling of the parasite between its two hosts.

The parasite Trypanosoma brucei cycles between an insect and a mammalian host and is the causative agent of sleeping sickness. Here, we performed high-throughput mapping of pseudouridines (Ψs) on mRNA from two life stages of the parasite. The analysis revealed ~273 Ψs, including developmentally regulated Ψs that are guided by homologs of pseudouridine synthases (PUS1, 3, 5, and 7). Mutating the U that undergoes pseudouridylation in the 3' UTR of valyl-tRNA synthetase destabilized the mRNA level. To investigate the mechanism by which Ψ affects the stability of this mRNA, proteins that bind to the 3' UTR were identified, including the RNA binding protein RBSR1. The binding of RBSR1 protein to the 3' UTR was stronger when lacking Ψ compared to transcripts carrying the modification, suggesting that Ψ can inhibit the binding of proteins to their target and thus affect the stability of mRNAs. Consequently, Ψ modification on mRNA adds an additional level of regulation to the dominant post-transcriptional control in these parasites.

Blood pressure (BP) is determined by several physiological factors that are regulated by a range of complex neural, endocrine, and paracrine mechanisms. This study examined a collection of 198 human genes related to BP regulation, in the biological processes and functional prisms, as well as gene expression in organs and tissues. This was made in conjunction with an orthology analysis performed in 19 target organisms along the phylogenetic tree. We have demonstrated that transport and signaling, as well as homeostasis in general, are the most prevalent biological processes associated with BP gene orthologs across the examined species. We showed that these genes and their orthologs are expressed primarily in the kidney and adrenals of complex organisms (e.g., high order vertebrates) and in the nervous system of low complexity organisms (e.g., flies, nematodes). Furthermore, we have determined that basic functions such as ion transport are ancient and appear in all organisms, while more complex regulatory functions, such as control of extracellular volume emerged in high order organisms. Thus, we conclude that the complex system of BP regulation evolved from simpler components that were utilized to maintain specific homeostatic functions that play key roles in existence and survival of organisms.