Accurate structure prediction of biomolecular interactions with AlphaFold 3.

The introduction of AlphaFold 21 has spurred a revolution in modelling the structure of proteins and their interactions, enabling a huge range of applications in protein modelling and design2-6. Here we describe our AlphaFold 3 model with a substantially updated diffusion-based architecture that is capable of predicting the joint structure of complexes including proteins, nucleic acids, small molecules, ions and modified residues. The new AlphaFold model demonstrates substantially improved accuracy over many previous specialized tools: far greater accuracy for protein-ligand interactions compared with state-of-the-art docking tools, much higher accuracy for protein-nucleic acid interactions compared with nucleic-acid-specific predictors and substantially higher antibody-antigen prediction accuracy compared with AlphaFold-Multimer v.2.37,8. Together, these results show that high-accuracy modelling across biomolecular space is possible within a single unified deep-learning framework.

Highly accurate protein structure prediction with AlphaFold.

Proteins are essential to life, and understanding their structure can facilitate a mechanistic understanding of their function. Through an enormous experimental effort1-4, the structures of around 100,000 unique proteins have been determined5, but this represents a small fraction of the billions of known protein sequences6,7. Structural coverage is bottlenecked by the months to years of painstaking effort required to determine a single protein structure. Accurate computational approaches are needed to address this gap and to enable large-scale structural bioinformatics. Predicting the three-dimensional structure that a protein will adopt based solely on its amino acid sequence-the structure prediction component of the 'protein folding problem'8-has been an important open research problem for more than 50 years9. Despite recent progress10-14, existing methods fall far short of atomic accuracy, especially when no homologous structure is available. Here we provide the first computational method that can regularly predict protein structures with atomic accuracy even in cases in which no similar structure is known. We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Critical Assessment of protein Structure Prediction (CASP14)15, demonstrating accuracy competitive with experimental structures in a majority of cases and greatly outperforming other methods. Underpinning the latest version of AlphaFold is a novel machine learning approach that incorporates physical and biological knowledge about protein structure, leveraging multi-sequence alignments, into the design of the deep learning algorithm.

Highly accurate protein structure prediction for the human proteome.

Protein structures can provide invaluable information, both for reasoning about biological processes and for enabling interventions such as structure-based drug development or targeted mutagenesis. After decades of effort, 17% of the total residues in human protein sequences are covered by an experimentally determined structure1. Here we markedly expand the structural coverage of the proteome by applying the state-of-the-art machine learning method, AlphaFold2, at a scale that covers almost the entire human proteome (98.5% of human proteins). The resulting dataset covers 58% of residues with a confident prediction, of which a subset (36% of all residues) have very high confidence. We introduce several metrics developed by building on the AlphaFold model and use them to interpret the dataset, identifying strong multi-domain predictions as well as regions that are likely to be disordered. Finally, we provide some case studies to illustrate how high-quality predictions could be used to generate biological hypotheses. We are making our predictions freely available to the community and anticipate that routine large-scale and high-accuracy structure prediction will become an important tool that will allow new questions to be addressed from a structural perspective.

Improved protein structure prediction using potentials from deep learning.

Protein structure prediction can be used to determine the three-dimensional shape of a protein from its amino acid sequence1. This problem is of fundamental importance as the structure of a protein largely determines its function2; however, protein structures can be difficult to determine experimentally. Considerable progress has recently been made by leveraging genetic information. It is possible to infer which amino acid residues are in contact by analysing covariation in homologous sequences, which aids in the prediction of protein structures3. Here we show that we can train a neural network to make accurate predictions of the distances between pairs of residues, which convey more information about the structure than contact predictions. Using this information, we construct a potential of mean force4 that can accurately describe the shape of a protein. We find that the resulting potential can be optimized by a simple gradient descent algorithm to generate structures without complex sampling procedures. The resulting system, named AlphaFold, achieves high accuracy, even for sequences with fewer homologous sequences. In the recent Critical Assessment of Protein Structure Prediction5 (CASP13)-a blind assessment of the state of the field-AlphaFold created high-accuracy structures (with template modelling (TM) scores6 of 0.7 or higher) for 24 out of 43 free modelling domains, whereas the next best method, which used sampling and contact information, achieved such accuracy for only 14 out of 43 domains. AlphaFold represents a considerable advance in protein-structure prediction. We expect this increased accuracy to enable insights into the function and malfunction of proteins, especially in cases for which no structures for homologous proteins have been experimentally determined7.

AlphaFold Protein Structure Database in 2024: providing structure coverage for over 214 million protein sequences.

The AlphaFold Database Protein Structure Database (AlphaFold DB, https://alphafold.ebi.ac.uk) has significantly impacted structural biology by amassing over 214 million predicted protein structures, expanding from the initial 300k structures released in 2021. Enabled by the groundbreaking AlphaFold2 artificial intelligence (AI) system, the predictions archived in AlphaFold DB have been integrated into primary data resources such as PDB, UniProt, Ensembl, InterPro and MobiDB. Our manuscript details subsequent enhancements in data archiving, covering successive releases encompassing model organisms, global health proteomes, Swiss-Prot integration, and a host of curated protein datasets. We detail the data access mechanisms of AlphaFold DB, from direct file access via FTP to advanced queries using Google Cloud Public Datasets and the programmatic access endpoints of the database. We also discuss the improvements and services added since its initial release, including enhancements to the Predicted Aligned Error viewer, customisation options for the 3D viewer, and improvements in the search engine of AlphaFold DB.

The AlphaFold Protein Structure Database (AlphaFold DB) is a massive digital library of predicted protein structures, with over 214 million entries, marking a 500-times expansion in size since its initial release in 2021. The structures are predicted using Google DeepMind's AlphaFold 2 artificial intelligence (AI) system. Our new report highlights the latest updates we have made to this database. We have added more data on specific organisms and proteins related to global health and expanded to cover almost the complete UniProt database, a primary data resource of protein sequences. We also made it easier for our users to access the data by directly downloading files or using advanced cloud-based tools. Finally, we have also improved how users view and search through these protein structures, making the user experience smoother and more informative. In short, AlphaFold DB has been growing rapidly and has become more user-friendly and robust to support the broader scientific community.

Too Much Too Little: Clarifying the Relationship Between Maternal Iodine Intake and Neurodevelopmental Outcomes.

BACKGROUND: In 2009, the Australian government mandated the fortification of bread salt with iodine. In 2010, pregnant and lactating women were also advised to take an iodine-containing supplement. Our assessment of this policy in an iodine-sufficient population showed that children whose mothers were in the highest and lowest quartiles of iodine intake performed more poorly on early childhood tests of cognition and language than those in the second quartile. However, we did not quantify the iodine intake associated with optimal neurodevelopment. OBJECTIVES: The aim was to establish the iodine intake range in pregnancy associated with optimal child neurodevelopment. METHODS: A prospective cohort study of pregnant women and their young children (n = 699). Iodine intake was assessed by a validated food frequency questionnaire at 16 and 28 wk of gestation. Child neurodevelopment at 18 mo of age was measured using the Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III). The relationship between average iodine intake during pregnancy and child neurodevelopment was assessed using linear regression with fractional polynomials and adjustment for confounders. RESULTS: Mean (SD) iodine intake was similar at study entry and 28 wk, 308 (120) µg/d, with 82% of women taking iodine supplements at study entry. The relationship between iodine intake during pregnancy and Bayley-III cognitive and language scores was curvilinear (P = 0.001 and P = 0.004, respectively), with the lowest Bayley-III scores observed at lower and higher iodine intakes. The inflection point that drove the association between lower iodine intake in pregnancy and poorer child neurodevelopment scores was around 185 µg/d; for the higher pregnancy iodine intakes, language and cognitive scores were negatively affected from ∼350 µg/d to 370 µg/d, respectively. Higher iodine intakes were being driven by supplement use. CONCLUSIONS: Targeted, not blanket, iodine supplementation may be needed for pregnant women with low-iodine intake from food.

The Effect of Iron Supplements on the Gut Microbiome of Females of Reproductive Age: A Randomized Controlled Trial.

BACKGROUND: Iron deficiency is the most common nutritional deficiency worldwide, particularly for young children and females of reproductive age. Although oral iron supplements are routinely recommended and generally considered safe, iron supplementation has been shown to alter the fecal microbiota in low-income countries. Little is known about the effect of iron supplementation on the fecal microbiota in high-income settings. OBJECTIVES: To assess the effect of oral iron supplementation compared with placebo on the gut microbiome in nonpregnant females of reproductive age in a high-income country. METHODS: A 21-d prospective parallel design double-blind, randomized control trial conducted in South Australia, Australia. Females (18-45 y) were randomly assigned to either iron (65.7 mg ferrous fumarate) or placebo. Fecal samples were collected prior to commencing supplements and after 21 d of supplementation. The primary outcome was microbiota ß-diversity (paired-sample weighted unique fraction metric dissimilarity) between treatment and placebo groups after 21 d of supplementation. Exploratory outcomes included changes in the relative abundance of bacterial taxa. RESULTS: Of 82 females randomly assigned, 80 completed the trial. There was no significant difference between the groups for weighted unique fraction metric dissimilarity (mean difference: 0.003; 95% confidence interval: -0.007, 0.014; P = 0.52) or relative abundance of common bacterial taxa or Escherichia-Shigella (q > 0.05). CONCLUSIONS: Iron supplementation did not affect the microbiome of nonpregnant females of reproductive age in Australia. This trial was registered at clinicaltrials.gov as NCT05033483.

Urinary Ferritin as a Noninvasive Means of Assessing Iron Status in Young Children.

BACKGROUND: Iron deficiency (ID) is the most common nutritional deficiency affecting young children. Serum ferritin concentration is the preferred biomarker for measuring iron status because it reflects iron stores; however, blood collection can be distressing for young children and can be logistically difficult. A noninvasive means to measure iron status would be attractive to either diagnose or screen for ID in young children. OBJECTIVES: This study aimed to determine the correlation between urinary and serum ferritin concentrations in young children; to determine whether correcting urinary ferritin for creatinine and specific gravity improves the correlation; and to determine a urine ferritin cut point to predict ID. METHODS: Validation study was conducted using paired serum and urine collected from 3-y-old children (n = 142) participating in a longitudinal birth cohort study: the ORIGINS project in Perth, Western Australia. We calculated the sensitivity, specificity, positive, and negative predictive values of urinary ferritin amount in identifying those with ID at the clinical cut point used by the World Health Organization (serum ferritin concentration of <12 ng/mL). RESULTS: Urine ferritin, corrected for creatinine, correlated moderately with serum ferritin [r = 0.53 (0.40-0.64)] and performed well in predicting those with ID (area under the curve: 0.85; 95% confidence interval: 0.75, 0.94). Urine ferritin <2.28 ng/mg creatinine was sensitive (86%) and specific (77%) in predicting ID and had a high negative predictive value of 97%; however, the positive predictive value was low (40%) owing to the low prevalence of ID in the sample (16%). CONCLUSIONS: Urine ferritin shows good diagnostic performance for ID. This noninvasive biomarker maybe a useful screening tool to exclude ID in healthy young children; however, further research is needed in other populations.

AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models.

The AlphaFold Protein Structure Database (AlphaFold DB, https://alphafold.ebi.ac.uk) is an openly accessible, extensive database of high-accuracy protein-structure predictions. Powered by AlphaFold v2.0 of DeepMind, it has enabled an unprecedented expansion of the structural coverage of the known protein-sequence space. AlphaFold DB provides programmatic access to and interactive visualization of predicted atomic coordinates, per-residue and pairwise model-confidence estimates and predicted aligned errors. The initial release of AlphaFold DB contains over 360,000 predicted structures across 21 model-organism proteomes, which will soon be expanded to cover most of the (over 100 million) representative sequences from the UniRef90 data set.

Iron-Fortified Foods Are Needed To Meet the Estimated Average Requirement for Iron in Australian Infants Aged 6 to 12 Months.

BACKGROUND: Meeting iron intake recommendations is challenging for infants 6-12 mo, especially breastfed infants. Three-quarters of Australian infants 6-12 mo have iron intakes below the estimated average requirement (7 mg), placing them at risk of iron deficiency. After 6 mo, breastmilk is no longer sufficient to meet the increased demand for iron, and iron-rich complementary foods are recommended. Iron-fortified foods may be a means of improving iron intake in infants, particularly those that are breastfed. OBJECTIVES: The aims of the study were as follows: 1) to examine the effect of milk-type and fortified foods on iron intake and the prevalence of inadequacy in infants 6-12 mo; 2) to model the effect of fixed amounts of iron-fortified infant cereal (IFIC) at 6 levels of iron fortification on total iron intake and the prevalence of inadequacy; and 3) to assess the effect IFIC on the intake of other nutrients in the diet. DESIGN: Secondary analysis of cross-sectional dietary intake data of infants 6-12 mo (n = 286) participating in the Australian Feeding Infants and Toddlers Study (OzFITS) 2021. RESULTS: Median (interquartile range) iron intake was 8.9 (7.5, 10.3); 6.3 (4.5, 8.2); and 2.7 (1.5, 4.4) mg/d in formula-fed, combination-fed, and breastfed infants, respectively. The corresponding prevalence of inadequacy was 19%, 67%, and 96%. Infants who consumed fortified foods had higher median iron intakes than those who did not, 6.2 compared with 1.9 mg/d. Dietary modeling showed that consuming 18 g (300 kJ) of IFIC, fortified at 35 mg/100 g dry weight, reduces the prevalence of inadequacy for iron from 75% to 5% for all infants. CONCLUSIONS: Iron intakes are low in Australian infants, especially for breastfed infants in the second half of infancy. Modeling shows that 300 kJ of IFIC, the current manufacturer-recommended serving, fortified at 35 mg/100 g dry weight, added to infant diets would be an effective means to reduce the prevalence of inadequacy for iron.

Evaluation of Calibration Approaches for Indoor Deployments of PurpleAir Monitors.

Low-cost air quality monitors are growing in popularity among both researchers and community members to understand variability in pollutant concentrations. Several studies have produced calibration approaches for these sensors for ambient air. These calibrations have been shown to depend primarily on relative humidity, particle size distribution, and particle composition, which may be different in indoor environments. However, despite the fact that most people spend the majority of their time indoors, little is known about the accuracy of commonly used devices indoors. This stems from the fact that calibration data for sensors operating in indoor environments are rare. In this study, we sought to evaluate the accuracy of the raw data from PurpleAir fine particulate matter monitors and for published calibration approaches that vary in complexity, ranging from simply applying linear corrections to those requiring co-locating a filter sample for correction with a gravimetric concentration during a baseline visit. Our data includes PurpleAir devices that were co-located in each home with a gravimetric sample for 1-week periods (265 samples from 151 homes). Weekly-averaged gravimetric concentrations ranged between the limit of detection (3 µg/m3) and 330 µg/m3. We found a strong correlation between the PurpleAir monitor and the gravimetric concentration (R>0.91) using internal calibrations provided by the manufacturer. However, the PurpleAir data substantially overestimated indoor concentrations compared to the gravimetric concentration (mean bias error ≥ 23.6 µg/m3 using internal calibrations provided by the manufacturer). Calibrations based on ambient air data maintained high correlations (R ≥ 0.92) and substantially reduced bias (e.g. mean bias error = 10.1 µg/m3 using a US-wide calibration approach). Using a gravimetric sample from a baseline visit to calibrate data for later visits led to an improvement over the internal calibrations, but performed worse than the simpler calibration approaches based on ambient air pollution data. Furthermore, calibrations based on ambient air pollution data performed best when weekly-averaged concentrations did not exceed 30 µg/m3, likely because the majority of the data used to train these models were below this concentration.

Restricting diet for perceived health benefit: A mixed-methods exploration of peripartum food taboos in rural Cambodia.

Food taboos encompass food restrictions practiced by a group that go beyond individual preferences. During pregnancy and lactation, food taboos may contribute to inadequate nutrition and poor maternal and infant health. Restriction of specific fish, meat, fruits and vegetables is common among peripartum women in many Southeast Asian countries, but data from Cambodia are lacking. In this mixed-methods study, 335 Cambodian mothers were asked open-ended questions regarding dietary behaviours during pregnancy and up to 24 weeks postpartum. Descriptive statistics and content analysis were used to characterize food taboos and multiple logistic regression analyses were conducted to identify predictors of this practice. Participants were 18-44 years of age, all of Khmer ethnicity and 31% were primiparous. Sixty-six per cent of women followed food taboos during the first 2 weeks postpartum, whereas ~20% of women restricted foods during other peripartum periods. Pregnancy taboos were often beneficial, including avoidance of sugar-sweetened beverages, coffee and alcohol. Conversely, postpartum avoidances typically included nutrient-dense foods such as fish, raw vegetables and chicken. Food taboos were generally followed to support maternal and child health. No significant predictors of food taboos during pregnancy were identified. Postpartum, each additional live birth a woman had reduced her odds of following food taboos by 24% (odds ratio [95% confidence interval]: 0.76 [0.61-0.95]). Specific food taboo practices and rationales varied greatly between women, suggesting that food taboos are shaped less by a strict belief system within the Khmer culture and more by individual or household understandings of food and health during pregnancy and postpartum.

Applying and improving AlphaFold at CASP14.

We describe the operation and improvement of AlphaFold, the system that was entered by the team AlphaFold2 to the "human" category in the 14th Critical Assessment of Protein Structure Prediction (CASP14). The AlphaFold system entered in CASP14 is entirely different to the one entered in CASP13. It used a novel end-to-end deep neural network trained to produce protein structures from amino acid sequence, multiple sequence alignments, and homologous proteins. In the assessors' ranking by summed z scores (>2.0), AlphaFold scored 244.0 compared to 90.8 by the next best group. The predictions made by AlphaFold had a median domain GDT_TS of 92.4; this is the first time that this level of average accuracy has been achieved during CASP, especially on the more difficult Free Modeling targets, and represents a significant improvement in the state of the art in protein structure prediction. We reported how AlphaFold was run as a human team during CASP14 and improved such that it now achieves an equivalent level of performance without intervention, opening the door to highly accurate large-scale structure prediction.

Baseline Hemoglobin, Hepcidin, Ferritin, and Total Body Iron Stores are Equally Strong Diagnostic Predictors of a Hemoglobin Response to 12 Weeks of Daily Iron Supplementation in Cambodian Women.

BACKGROUND: The WHO recommends daily iron supplementation for all women in areas where the population-level anemia prevalence is ≥40%, despite the fact that hemoglobin (Hb) concentration is generally considered to be a poor prognostic indicator of iron status. OBJECTIVES: In this secondary analysis, we investigated the predictive power of ten baseline hematological biomarkers towards a 12-week Hb response to iron supplementation. METHODS: Data were obtained from a randomized controlled trial of daily iron supplementation in 407 nonpregnant Cambodian women (18-45 years) who received 60 mg elemental iron as ferrous sulfate for 12 weeks. Ten baseline biomarkers were included: Hb, measured with both a hematology analyzer and a HemoCue; inflammation-adjusted ferritin; soluble transferrin receptor; reticulocyte Hb; hepcidin; mean corpuscular volume; inflammation-adjusted total body iron stores (TBIS); total iron binding capacity; and transferrin saturation. Receiver operating characteristic (ROC) curves from fitted logistic regression models were used to make discrimination comparisons and variable selection methods were used to construct a multibiomarker prognostic model. RESULTS: Only 25% (n = 95/383) of women who completed the trial experienced a 12-week Hb response ≥10 g/L. The strongest univariate predictors of a Hb response were Hb as measured with a hematology analyzer, inflammation-adjusted ferritin, hepcidin, and inflammation-adjusted TBIS (AUCROC = 0.81, 0.83, 0.82, and 0.82, respectively), and the optimal cutoffs to identify women who were likely to experience a Hb response were 117 g/L, 17.3 µg/L, 1.98 nmol/L, and 1.95 mg/kg, respectively. Hb as measured with a hematology analyzer, inflammation-adjusted ferritin, and hepcidin had the best combined predictive ability (AUCROC=0.86). Hb measured with the HemoCue had poor discrimination ability (AUCROC = 0.65). CONCLUSIONS: Baseline Hb as measured with a hematology analyzer was as strong a predictor of Hb response to iron supplementation as inflammation-adjusted ferritin, hepcidin, and inflammation-adjusted TBIS. This is positive given that the WHO currently uses the population-level anemia prevalence to guide recommendations for untargeted iron supplementation.

Maternal Late-Pregnancy Serum Unmetabolized Folic Acid Concentrations Are Not Associated with Infant Allergic Disease: A Prospective Cohort Study.

BACKGROUND: The increase in childhood allergic disease in recent decades has coincided with increased folic acid intakes during pregnancy. Circulating unmetabolized folic acid (UMFA) has been proposed as a biomarker of excessive folic acid intake. OBJECTIVE: We aimed to determine if late-pregnancy serum UMFA and total folate concentrations were associated with allergic disease risk in the offspring at 1 y of age in a population at high risk of allergy. METHODS: The cohort consisted of 561 mother-infant pairs from Western Australia. To be eligible the infant had to have a first-degree relative (mother, father, or sibling) with a history of medically diagnosed allergic disease. Maternal venous blood was collected between 36 and 40 wk of gestation. Serum UMFA was measured by LC-tandem MS. Serum total folate was determined using a microbiological method with chloramphenicol-resistant Lactobacillus rhamnosus as the test organism, and was collected between 36 and 40 wk of gestation. UMFA concentrations were measured by tandem MS using stable isotope dilution; folate concentrations were determined using the microbiological method with standardized kits. Infant allergic disease outcomes of medically diagnosed eczema, steroid-treated eczema, atopic eczema, IgE-mediated food allergy, allergen sensitization, and medically diagnosed wheeze were assessed at 1 y of age. RESULTS: Median (IQR) concentrations for UMFA and serum folate were 1.6 (0.6-4.7) and 53.2 (32.6-74.5) nmol/L, respectively. Of the infants, 34.6% had medically diagnosed eczema, 26.4% allergen sensitization, and 14.9% had an IgE-mediated food allergy. In both adjusted and unadjusted models there was little evidence of association between UMFA or serum folate and any of the infant allergy outcomes. CONCLUSIONS: In this cohort of children at high risk of allergic disease there was no association between maternal UMFA or serum folate concentrations measured in late pregnancy and allergic disease outcomes at 1 y of age.

Effect of a functional fibre supplement on glycemic control when added to a year-long medically supervised weight management program in adults with type 2 diabetes.

PURPOSE: Soluble fibre beneficially affects metabolism but whether it can augment the reductions in glycemia induced through intensive weight management has not been extensively studied. Our objective was to examine the adjunct effect of the soluble viscous fibre PGX® on glycemic control in adults with type 2 diabetes (T2D) enrolled in a year-long medically supervised weight management program. METHODS: In a placebo-controlled, double-blind study, 290 adults with overweight/obesity and T2D were randomized to receive PGX (15-20 g/day) or isocaloric placebo (rice flour, 6.4-8.6 g/day) as an adjunct to intensive weight management for 52 weeks. The primary outcome was change in glycemic control (HbA1c). Other outcome measures included weight loss, blood lipids, blood pressure, cytokines and fecal microbiota. RESULTS: Compared to baseline HbA1c in PGX (7.2 ± 1.1%) and placebo (7.0 ± 0.9%) groups, there was a significant reduction at 16 and 26 weeks, however, only PGX showed a significant absolute reduction of 0.23% at 52 weeks; there were no between-group differences in HbA1c. At 52 weeks, only PGX significantly decreased body weight compared to baseline and reduced waist circumference at all time points. Compared to baseline, only PGX showed a significant reduction in LDL cholesterol at 16 and 26 weeks. PGX significantly increased the relative abundance of Collinsella, Parabacteroides and Roseburia. CONCLUSION: Adding PGX to a weight management program for individuals with T2D provides a sustained reduction in HbA1c compared to placebo. Improvements in other metabolic outcomes suggest that PGX may be a promising adjunct to weight loss programs in patients with T2D. CLINICAL TRIAL: This trial was registered at ClinicalTrials.gov as NCT01644201.

The Homozygous Hemoglobin EE Variant Is Associated with Poorer Riboflavin Status in Cambodian Women of Reproductive Age.

BACKGROUND: Riboflavin is required for erythropoiesis, which is increased in people with hemoglobinopathies due to increased hemolysis and erythrocyte turnover. Dietary intake and status of riboflavin is poor in Cambodia, where hemoglobinopathies are common. OBJECTIVE: We assessed the association between genetic hemoglobin disorders and riboflavin status in women of reproductive age in Cambodia. METHODS: Venous blood samples from 515 Cambodian women of reproductive age, 18-45 y, were analyzed for biomarker status of riboflavin [erythrocyte glutathione reductase activation coefficient (EGRac)], genetic hemoglobin (Hb) disorders, and hematological indices. Linear regression analysis was used to estimate the association between EGRac with Hb, ferritin, and Hb genotypes. EGRac was log transformed in the analyses, and the regression coefficients represent the geometric mean differences. RESULTS: Genetic Hb disorders were present in 57% of the population, with the homozygous hemoglobin E variant (Hb EE) occurring in ∼10% of women (n = 53). Deficient (EGRac ≥1.40) or marginal riboflavin status (EGRac ≥1.30 and <1.40) was observed in 92% (n = 475) of women. The variant Hb EE genotype was associated with 18% (95% CI: 9%, 28%) higher geometric mean EGRac values than the normal Hb AA genotype (P < 0.001). CONCLUSIONS: Although riboflavin biomarker deficiency or marginal status is widely prevalent in Cambodian women, lower riboflavin status was observed more frequently in women with the Hb EE genotype than in women with normal Hb AA. The relation between genetic Hb disorders and riboflavin warrants further investigation. This trial was registered at clinicaltrials.gov as NCT01593423 and NCT02481375.

Are the nutrient and textural properties of Australian commercial infant and toddler foods consistent with infant feeding advice?

Infant feeding guidelines worldwide recommend first foods to be Fe rich with no added sugars and that nutrient-poor discretionary foods are to be avoided. Feeding guidelines also recommend exposing infants to a variety of foods and flavours with increasingly complex textures. Here, we compare nutritional and textural properties of commercial infant and toddler foods available in Australia with established infant feeding guidelines. Nutrition information and ingredient lists were obtained from food labels, manufacturer and/or retailer websites. In total, 414 foods were identified, comprising mostly mixed main dishes, fruit and vegetable first foods and snacks. Most products were poor sources of Fe, and 80 % of first foods were fruit-based. Half of all products were purées in squeeze pouches, and one-third of all products were discretionary foods. The nutritional content of many products was inconsistent with guidelines, being low in Fe, sweet, smooth in consistency or classified as discretionary. Reformulation of products is warranted to improve Fe content, particularly in mixed main dishes, expand the range of vegetable-only foods and textural variety. Greater regulatory oversight may be needed to better inform parents and caregivers. Frequent consumption of commercial baby foods low in Fe may increase the risk of Fe deficiency. Excessive consumption of purées via squeeze pouches may also have implications for overweight and obesity risk.

Transformative instruction or old wine in new skins? Exploring how and why educators use HyperDocs.

HyperDocs are interactive, digital teaching and learning materials created, disseminated, and remixed by educators. To date, HyperDocs have not been the subject of published, peer-reviewed research. To address this research gap, we engaged in exploratory, primarily qualitative research to systematically examine how and why teachers use HyperDocs. We used an online survey to gather data on educators' (N = 261) uses of and perceptions regarding HyperDocs. Analysis suggested a wide range of definitions of, purposes for, and approaches to HyperDoc use, indicating that educators are adapting HyperDocs to their practice in myriad ways. Consistent with the openness and flexibility in finding, remixing, and using HyperDocs, educators identified a number of benefits of using these tools in their practice, including changes in student engagement and learning, shifts in instructional design and delivery, and changes in their own support and dispositions. Analysis of examples of HyperDocs shared by a subset of participants suggested some mismatch between rhetoric about HyperDocs and what was actually incorporated into them. We discuss these findings in relation to the work of educators and the future of research on HyperDocs and other crowdsourced teaching and learning initiatives.

Protein structure prediction using multiple deep neural networks in the 13th Critical Assessment of Protein Structure Prediction (CASP13).

We describe AlphaFold, the protein structure prediction system that was entered by the group A7D in CASP13. Submissions were made by three free-modeling (FM) methods which combine the predictions of three neural networks. All three systems were guided by predictions of distances between pairs of residues produced by a neural network. Two systems assembled fragments produced by a generative neural network, one using scores from a network trained to regress GDT_TS. The third system shows that simple gradient descent on a properly constructed potential is able to perform on par with more expensive traditional search techniques and without requiring domain segmentation. In the CASP13 FM assessors' ranking by summed z-scores, this system scored highest with 68.3 vs 48.2 for the next closest group (an average GDT_TS of 61.4). The system produced high-accuracy structures (with GDT_TS scores of 70 or higher) for 11 out of 43 FM domains. Despite not explicitly using template information, the results in the template category were comparable to the best performing template-based methods.