Background: Postpartum hemorrhage remains one of the largest causes of maternal morbidity and mortality in the United States. Objective: The aim of this paper is to use machine learning techniques to identify patients at risk for postpartum hemorrhage at obstetric delivery. Methods: Women aged 18 to 55 years delivering at a major academic center from July 2013 to October 2018 were included for analysis (N=30,867). A total of 497 variables were collected from the electronic medical record including the following: demographic information; obstetric, medical, surgical, and family history; vital signs; laboratory results; labor medication exposures; and delivery outcomes. Postpartum hemorrhage was defined as a blood loss of ≥1000 mL at the time of delivery, regardless of delivery method, with 2179 (7.1%) positive cases observed. Supervised learning with regression-, tree-, and kernel-based machine learning methods was used to create classification models based upon training (21,606/30,867, 70%) and validation (4630/30,867, 15%) cohorts. Models were tuned using feature selection algorithms and domain knowledge. An independent test cohort (4631/30,867, 15%) determined final performance by assessing for accuracy, area under the receiver operating curve (AUROC), and sensitivity for proper classification of postpartum hemorrhage. Separate models were created using all collected data versus models limited to data available prior to the second stage of labor or at the time of decision to proceed with cesarean delivery. Additional models examined patients by mode of delivery. Results: Gradient boosted decision trees achieved the best discrimination in the overall model. The model including all data mildly outperformed the second stage model (AUROC 0.979, 95% CI 0.971-0.986 vs AUROC 0.955, 95% CI 0.939-0.970). Optimal model accuracy was 98.1% with a sensitivity of 0.763 for positive prediction of postpartum hemorrhage. The second stage model achieved an accuracy of 98.0% with a sensitivity of 0.737. Other selected algorithms returned models that performed with decreased discrimination. Models stratified by mode of delivery achieved good to excellent discrimination but lacked the sensitivity necessary for clinical applicability. Conclusions: Machine learning methods can be used to identify women at risk for postpartum hemorrhage who may benefit from individualized preventative measures. Models limited to data available prior to delivery perform nearly as well as those with more complete data sets, supporting their potential utility in the clinical setting. Further work is necessary to create successful models based upon mode of delivery and to validate the findings of this study. An unbiased approach to hemorrhage risk prediction may be superior to human risk assessment and represents an area for future research.

Retrotransposons are genomic DNA sequences that are capable of copying themselves to new genomic locations via RNA intermediates; LINE-1 is the only retrotransposon that remains autonomous and active in the human genome. The mobility of LINE-1 is largely repressed in somatic tissues, but LINE-1 is active in many cancers. Recent studies using LINE-1 constructs indicate that host cells activate a DNA damage response (DDR) to repair retrotransposition intermediates and resolve conflicts between LINE-1 and DNA replication. Using multi-omic data from the CPTAC project, we found correlations between LINE-1 expression and ATM-MRN-SMC DDR signalling in endometrial cancer and between LINE-1 and the ATR-CHEK1 pathway in p53 wild type breast cancer. This provides evidence that conflicts between LINE-1 and DNA replication occur in at least some human cancers. Furthermore, LINE-1 expression in these cancers is correlated with the total amount of copy number variation genome wide, indicating that, when active in cancer, pointing to a direct impact of LINE-1 associated DNA damage on genome structure. We also find that, in endometrial and ovarian cancer, LINE-1 expression is correlated with the expression of genes that drive cycle progression including E2F3, PLK1 and Aurora kinase B. This study provides evidence, supporting recent work in model cell lines, of a LINE-1/DDR connection in human tumors and raises the possibility of additional interactions between LINE-1 and the cell cycle.

Background: Postpartum hemorrhage remains one of the largest causes of maternal morbidity and mortality in the United States. Objective: To utilize machine learning techniques to identify patients at risk for postpartum hemorrhage at obstetric delivery. Study Design: Women aged 18 to 55 delivering at a major academic center from July 2013 to October 2018 were included for analysis (n = 30,867). A total of 497 variables were collected from the electronic medical record including demographic information, obstetric, medical, surgical, and family history, vital signs, laboratory results, labor medication exposures, and delivery outcomes. Postpartum hemorrhage was defined as a blood loss of 1000 mL at the time of delivery, regardless of delivery method, with 2179 positive cases observed (7.06%). Supervised learning with regression-, tree-, and kernel-based machine learning methods was used to create classification models based upon training (n = 21,606) and validation (n = 4,630) cohorts. Models were tuned using feature selection algorithms and domain knowledge. An independent test cohort (n = 4,631) determined final performance by assessing for accuracy, area under the receiver operating curve (AUC), and sensitivity for proper classification of postpartum hemorrhage. Separate models were created using all collected data versus limited to data available prior to the second stage of labor/at the time of decision to proceed with cesarean delivery. Additional models examined patients by mode of delivery. Results: Gradient boosted decision trees achieved the best discrimination in the overall model. The model including all data mildly outperformed the second stage model (AUC 0.979, 95% CI 0.971–0.986 vs. AUC 0.955, 95% CI 0.939–0.970). Optimal model accuracy was 98.1% with a sensitivity of 0.763 for positive prediction of postpartum hemorrhage. The second stage model achieved an accuracy of 98.0% with a sensitivity of 0.737. Other selected algorithms returned ≥ models that performed with decreased discrimination. Models stratified by mode of delivery achieved good to excellent discrimination, but lacked sensitivity necessary for clinical applicability. Conclusions: Machine learning methods can be used to identify women at risk for postpartum hemorrhage who may benefit from individualized preventative measures. Models limited to data available prior to delivery perform nearly as well as those with more complete datasets, supporting their potential utility in the clinical setting. Further work is necessary to create successful models based upon mode of delivery. An unbiased approach to hemorrhage risk prediction may be superior to human risk assessment and represents an area for future research. Condensation: Machine learning methods can be successfully utilized to predict nearly three-quarters of women at risk of postpartum hemorrhage when undergoing obstetric delivery. AJOG at a Glance: Why was the study conducted? To determine patients at risk for postpartum hemorrhage using modern machine learning techniques on a robust data set directly derived from the electronic medical record What are the key findings? Using 28 predictor features, the model successfully classified 73.7% of patients who ultimately had a postpartum hemorrhage using information available prior to delivery Many previously identified risk factors for postpartum hemorrhage were not included in the final model, potentially discounting their contribution to hemorrhage risk Models stratified by delivery method achieved good to excellent discrimination but noted lower sensitivity and need further investigation What does this study add to what is already known? This study represents the largest cohort directly-derived from the electronic medical record to use machine learning techniques to identify patients at risk for postpartum hemorrhage

Mammalian metabolic engineering advances basic biology, bioproduction and cell therapy. However, as pathway complexity increases, so does the size of the combinatorial design space and required DNA constructs, rendering unbiased screens intractable. Here, we developed shotgun genetic engineering (SGE), which exploits the ease of delivering many barcoded small constructs—rather than a single large one—into mammalian cells. Each cell serves as an independent experiment, carrying a synthetic pathway that explores gene content, stoichiometry and organellar localization. Functional pathways are identified by sequencing barcodes from cells exhibiting the desired phenotype. Using SGE, we screened millions of pathway combinations to engineer essential amino acid biosynthesis, achieving near-wild-type growth without valine and enabling isoleucine-free growth in Chinese hamster ovary cells, as well as valine-free growth in Jurkat cells. Functional pathways favored mitochondrial localization and required the integration of tens of kilobases of synthetic DNA, beyond the scale of conventional screening. The resulting datasets support machine-learning-guided decoding and engineering of complex biosynthetic traits in mammalian systems.

Motivation: Retrotransposition is an important force in shaping the human genome and is involved in prenatal development, disease and aging. Current genome browsers are not optimized for visualizing the experimental evidence for retrotransposon insertions. Results: We have developed a specialized browser to visualize the evidence for retrotransposon insertions for both targeted and whole-genome sequencing data. Availability and implementation: TranspoScope’s source code, as well as installation instructions, are available at https://github.com/FenyoLab/transposcope.

Recent studies have revealed diverse amino acid, post-translational and non-canonical modifications of proteins in diverse organisms and tissues. However, their unbiased detection and analysis remain hindered by technical limitations. Here, we present a spectral alignment method for the identification of protein modifications using high-resolution mass spectrometry proteomics. Termed SAMPEI for Spectral Alignment-based Modified PEptide Identification, this open-source algorithm is designed for the discovery of functional protein and peptide signaling modifications, without prior knowledge of their identities. Using synthetic standards and controlled chemical labeling experiments, we demonstrate its high specificity and sensitivity for the discovery of sub-stoichiometric protein modifications in complex cellular extracts. SAMPEI mapping of mouse macrophage differentiation revealed diverse post-translational protein modifications, including distinct forms of cysteine itaconatylation. SAMPEI’s robust parameterization and versatility are expected to facilitate the discovery of biological modifications of diverse macromolecules. SAMPEI is available open-source from BioConda and GitHub.