For antibody sequence generative modeling, mixture models may be all you need.

MOTIVATION: Antibody therapeutic candidates must exhibit not only tight binding to their target but also good developability properties, especially low risk of immunogenicity. RESULTS: In this work, we fit a simple generative model, SAM, to sixty million human heavy and seventy million human light chains. We show that the probability of a sequence calculated by the model distinguishes human sequences from other species with the same or better accuracy on a variety of benchmark datasets containing >400 million sequences than any other model in the literature, outperforming large language models (LLMs) by large margins. SAM can humanize sequences, generate new sequences, and score sequences for humanness. It is both fast and fully interpretable. Our results highlight the importance of using simple models as baselines for protein engineering tasks. We additionally introduce a new tool for numbering antibody sequences which is orders of magnitude faster than existing tools in the literature. AVAILABILITY: All tools developed in this study are available at https://github.com/Wang-lab-UCSD/AntPack. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Linear-Scaling Kernels for Protein Sequences and Small Molecules Outperform Deep Learning While Providing Uncertainty Quantitation and Improved Interpretability.

Gaussian process (GP) is a Bayesian model which provides several advantages for regression tasks in machine learning such as reliable quantitation of uncertainty and improved interpretability. Their adoption has been precluded by their excessive computational cost and by the difficulty in adapting them for analyzing sequences (e.g., amino acid sequences) and graphs (e.g., small molecules). In this study, we introduce a group of random feature-approximated kernels for sequences and graphs that exhibit linear scaling with both the size of the training set and the size of the sequences or graphs. We incorporate these new kernels into our new Python library for GP regression, xGPR, and develop an efficient and scalable algorithm for fitting GPs equipped with these kernels to large datasets. We compare the performance of xGPR on 17 different benchmarks with both standard and state-of-the-art deep learning models and find that GP regression achieves highly competitive accuracy for these tasks while providing with well-calibrated uncertainty quantitation and improved interpretability. Finally, in a simple experiment, we illustrate how xGPR may be used as part of an active learning strategy to engineer a protein with a desired property in an automated way without human intervention.

The association between ventriculostomy - Related infection and clinical outcomes: A systematic review and meta-analysis.

BACKGROUND: Ventriculostomy - related infection (VRI) is a common complication of patients who require placement of an external ventricular drain (EVD). The clinical outcomes of people who are diagnosed with VRI is poorly characterised. We performed a systematic review and meta-analysis to assess the association between VRI, and clinical outcomes and resource use, in patients treated with an EVD. METHODS: We searched MEDLINE, EMBASE, CINAHL and the Cochrane Central Register of clinical trials to identify clinical trial and cohort studies that reported outcomes including mortality, functional outcome, duration of EVD insertion, and intensive care and hospital length of stay. Inclusion criteria and data extraction were conducted in duplicate. Where sufficient data were available, data synthesis was conducted using a random effects model to provide a pooled estimate of the association between VRI and clinical outcomes and resource use. We also pooled data to provide an estimate of the incidence of VRI in this population. RESULTS: Nineteen studies including 38,247 patients were included in the meta-analysis. There were twelve different definitions of VRI in the included studies. The pooled estimate of the incidence of VRI was 11 % (95 % confidence interval (CI), 9 % to 14 %). A diagnosis of VRI was not associated with an increase in the estimated odds ratio (OR) for mortality (OR 1.07, 95 % CI 0.59 to 1.92, p = 0.83 I2 = 83.5 %), nor was a diagnosis of VRI associated with changes in neurological outcome (OR 1.42, 95 % CI 0.36 to 5.56, p = 0.89, I2 = 0.3 %). Those diagnosed with VRI had longer intensive care unit length of stay (estimated pooled mean difference 8.4 days 95 % CI 3.4 to 13.4 days, p = 0.0009, I2 = 78.7 %) an increase in hospital length of stay (estimated mean difference 16.4 days. 95 % CI 11.6 to 21.2 days, p < 0.0005, I2 = 76.6 %), a prolonged duration of EVD placement (mean difference 5.24 days, 95 % CI 3.05 to 7.43, I2 = 78.2 %, p < 0.01), and an increased requirement for an internal ventricular shunt (OR 1.80, 95 % CI 1.32 to 2.46, I2 = 8.92 %, p < 0.01). CONCLUSIONS: Ventriculostomy related infection is not associated with increased mortality or an increased risk of poor neurological outcome, but is associated with prolonged duration of EVD placement, prolonged duration of ICU and hospital admission, and an increased rate of internal ventricular shunt placement.

The RESP AI model accelerates the identification of tight-binding antibodies.

High-affinity antibodies are often identified through directed evolution, which may require many iterations of mutagenesis and selection to find an optimal candidate. Deep learning techniques hold the potential to accelerate this process but the existing methods cannot provide the confidence interval or uncertainty needed to assess the reliability of the predictions. Here we present a pipeline called RESP for efficient identification of high affinity antibodies. We develop a learned representation trained on over 3 million human B-cell receptor sequences to encode antibody sequences. We then develop a variational Bayesian neural network to perform ordinal regression on a set of the directed evolution sequences binned by off-rate and quantify their likelihood to be tight binders against an antigen. Importantly, this model can assess sequences not present in the directed evolution library and thus greatly expand the search space to uncover the best sequences for experimental evaluation. We demonstrate the power of this pipeline by achieving a 17-fold improvement in the KD of the PD-L1 antibody Atezolizumab and this success illustrates the potential of RESP in facilitating general antibody development.

Engineering human JMJD2A tudor domains for an improved understanding of histone peptide recognition.

JMJD2A is a histone lysine demethylase which recognizes and demethylates H3K9me3 and H3K36me3 residues and is overexpressed in various cancers. It utilizes a tandem tudor domain to facilitate its own recruitment to histone sites, recognizing various di- and tri-methyl lysine residues with moderate affinity. In this study, we successfully engineered the tudor domain of JMJD2A to specifically bind to H4K20me3 with a 20-fold increase of affinity and improved selectivity. To reveal the molecular basis, we performed molecular dynamics and free energy decomposition analysis on the human JMJD2A tandem tudor domains bound to H4K20me2, H4K20me3, and H3K23me3 peptides to uncover the residues and conformational changes important for the enhanced binding affinity and selectivity toward H4K20me2/3. These analyses revealed new insights into understanding chromatin reader domains recognizing histone modifications and improving binding affinity and selectivity of these domains. Furthermore, we showed that the tight binding of JMJD2A to H4K20me2/3 is not sufficient to improve the efficiency of CRISPR-CAS9 mediated homology directed repair (HDR), suggesting a complicated relationship between JMJD2A and the DNA damage response beyond binding affinity toward the H4K20me2/3 mark.

The promise of graphene-based transistors for democratizing multiomics studies.

As biological research has synthesized genomics, proteomics, metabolomics, and transcriptomics into systems biology, a new multiomics approach to biological research has emerged. Today, multiomics studies are challenging and expensive. An experimental platform that could unify the multiple omics approaches to measurement could increase access to multiomics data by enabling more individual labs to successfully attempt multiomics studies. Field effect biosensing based on graphene transistors have gained significant attention as a potential unifying technology for such multiomics studies. This review article highlights the outstanding performance characteristics that makes graphene field effect transistor an attractive sensing platform for a wide variety of analytes important to system biology. In addition to many studies demonstrating the biosensing capabilities of graphene field effect transistors, they are uniquely suited to address the challenges of multiomics studies by providing an integrative multiplex platform for large scale manufacturing using the well-established processes of semiconductor industry. Furthermore, the resulting digital data is readily analyzable by machine learning to derive actionable biological insight to address the challenge of data compatibility for multiomics studies. A critical stage of systems biology will be democratizing multiomics study, and the graphene field effect transistor is uniquely positioned to serve as an accessible multiomics platform.

Selection of QPX7831, an Orally Bioavailable Prodrug of Boronic Acid ß-Lactamase Inhibitor QPX7728.

In recognition of the need for effective oral therapies to treat Gram-negative bacterial infections, efforts were directed toward identifying an oral prodrug of ß-lactamase inhibitor clinical candidate QPX7728. Seventeen prodrugs were synthesized; key properties investigated were rates of cleavage to the active form in vitro, pharmacokinetics across species, and crystallinity. Compound 5-Na (QPX7831 Sodium) emerged with optimal properties across all key attributes.

Rapid and Electronic Identification and Quantification of Age-Specific Circulating Exosomes via Biologically Activated Graphene Transistors.

Increasing access to modern clinical practices concomitantly extends lifespan, ironically revealing new classes of degenerative and inflammatory diseases of later years. Here, an electronic graphene field-effect transistor (gFET) is reported, termed EV-chip, for label-free, rapid identification and quantification of exosomes (EV) associated with aging through specific surface markers, CD63 and CD151. Studies suggest that blood-derived exosomes carry specific biomolecules that can be used toward diagnostic applications of age and health. However, to observe improvements in patient outcomes, earlier detection at the point-of-care (POC) is required. Unfortunately, conventional techniques and other electronic-based platforms for exosome sensing are burdensome and inept for the POC distinction of aged blood factors. It is shown that EV-chip can quantitatively detect purified exosomes from plasma, with a limit of detection (LOD) of 2 × 104 particles mL-1 and a limit of quantification (LOQ) of 6 × 104 particles mL-1 . The sensitivity and compact electronics of the EV-chip improves upon previously published electronic biosensors, making it ideal for a physician's office or a simple biological laboratory. The sensitivity, selectivity, and portability of the EV-chip demonstrate the potential of the biosensor as a powerful point-of-care diagnostic and prognostic tool for age-related diseases.

Discrimination of single-point mutations in unamplified genomic DNA via Cas9 immobilized on a graphene field-effect transistor.

Simple and fast methods for the detection of target genes with single-nucleotide specificity could open up genetic research and diagnostics beyond laboratory settings. We recently reported a biosensor for the electronic detection of unamplified target genes using liquid-gated graphene field-effect transistors employing an RNA-guided catalytically deactivated CRISPR-associated protein 9 (Cas9) anchored to a graphene monolayer. Here, using unamplified genomic samples from patients and by measuring multiple types of electrical response, we show that the biosensors can discriminate within one hour between wild-type and homozygous mutant alleles differing by a single nucleotide. We also show that biosensors using a guide RNA-Cas9 orthologue complex targeting genes within the protospacer-adjacent motif discriminated between homozygous and heterozygous DNA samples from patients with sickle cell disease, and that the biosensors can also be used to rapidly screen for guide RNA-Cas9 complexes that maximize gene-targeting efficiency.

Engineering a Histone Reader Protein by Combining Directed Evolution, Sequencing, and Neural Network Based Ordinal Regression.

Directed evolution is a powerful approach for engineering proteins with enhanced affinity or specificity for a ligand of interest but typically requires many rounds of screening/library mutagenesis to obtain mutants with desired properties. Furthermore, mutant libraries generally only cover a small fraction of the available sequence space. Here, for the first time, we use ordinal regression to model protein sequence data generated through successive rounds of sorting and amplification of a protein-ligand system. We show that the ordinal regression model trained on only two sorts successfully predicts chromodomain CBX1 mutants that would have stronger binding affinity with the H3K9me3 peptide. Furthermore, we can extract the predictive features using contextual regression, a method to interpret nonlinear models, which successfully guides identification of strong binders not even present in the original library. We have demonstrated the power of this approach by experimentally confirming that we were able to achieve the same improvement in binding affinity previously achieved through a more laborious directed evolution process. This study presents an approach that reduces the number of rounds of selection required to isolate strong binders and facilitates the identification of strong binders not present in the original library.

Discovery of Cyclic Boronic Acid QPX7728, an Ultrabroad-Spectrum Inhibitor of Serine and Metallo-ß-lactamases.

Despite major advances in the ß-lactamase inhibitor field, certain enzymes remain refractory to inhibition by agents recently introduced. Most important among these are the class B (metallo) enzyme NDM-1 of Enterobacteriaceae and the class D (OXA) enzymes of Acinetobacter baumannii. Continuing the boronic acid program that led to vaborbactam, efforts were directed toward expanding the spectrum to allow treatment of a wider range of organisms. Through key structural modifications of a bicyclic lead, stepwise gains in spectrum of inhibition were achieved, ultimately resulting in QPX7728 (35). This compound displays a remarkably broad spectrum of inhibition, including class B and class D enzymes, and is little affected by porin modifications and efflux. Compound 35 is a promising agent for use in combination with a ß-lactam antibiotic for the treatment of a wide range of multidrug resistant Gram-negative bacterial infections, by both intravenous and oral administration.

Activity of Meropenem-Vaborbactam in Mouse Models of Infection Due to KPC-Producing Carbapenem-Resistant Enterobacteriaceae.

Meropenem-vaborbactam (Vabomere) is highly active against Gram-negative pathogens, especially Klebsiella pneumoniae carbapenemase (KPC)-producing, carbapenem-resistant Enterobacteriaceae The objective of these studies was to evaluate the efficacy of meropenem alone and in combination with vaborbactam in mouse thigh and lung infection models. Thighs or lungs of neutropenic mice were infected with KPC-producing carbapenem-resistant Enterobacteriaceae, with meropenem MICs ranging from ≤0.06 to 8 mg/liter in the presence of 8 mg/liter vaborbactam. Mice were treated with meropenem alone or meropenem in combination with vaborbactam every 2 h for 24 h to provide exposures comparable to 2-g doses of each component in humans. Meropenem administered in combination with vaborbactam produced bacterial killing in all strains tested, while treatment with meropenem alone either produced less than 0.5 log CFU/tissue of bacterial killing or none at all. In the thigh model, 11 strains were treated with the combination of meropenem plus vaborbactam (300 plus 50 mg/kg of body weight). This combination produced from 0.8 to 2.89 logs of bacterial killing compared to untreated controls at the start of treatment. In the lung infection model, two strains were treated with the same dosage regimen of meropenem and vaborbactam. The combination produced more than 1.83 logs of bacterial killing against both strains tested compared to untreated controls at the start of treatment. Overall, these data suggest that meropenem-vaborbactam may have utility in the treatment of infections due to KPC-producing carbapenem-resistant Enterobacteriaceae.

Activity of Simulated Human Dosage Regimens of Meropenem and Vaborbactam against Carbapenem-Resistant Enterobacteriaceae in an In Vitro Hollow-Fiber Model.

The objective of these studies was to evaluate the exposures of meropenem and vaborbactam that would produce antibacterial activity and prevent resistance development in carbapenem-resistant Klebsiella pneumoniae carbapenemase (KPC)-producing Enterobacteriaceae strains when tested at an inoculum of 108 CFU/ml. Thirteen K. pneumoniae isolates, three Enterobacter cloacae isolates, and one Escherichia coli isolate were examined in an in vitro hollow-fiber model over 32 h. Simulated dosage regimens of 1 to 2 g of meropenem with 1 to 2 g of vaborbactam, with meropenem administered every 8 h by a 3-h infusion based on phase 1 or phase 3 patient pharmacokinetic data, were studied in the model. A dosage of 2 g of meropenem in combination with 2 g of vaborbactam was bactericidal against K. pneumoniae, E. cloacae, and E. coli strains, with meropenem-vaborbactam MICs of up to 8 mg/liter. When the vaborbactam exposure was adjusted to the levels observed in patients enrolled in phase 3 trials (24-h free AUC, ∼550 mg · h/liter, versus 320 mg · h/liter in the phase 1 studies), 2 g of meropenem with 2 g of vaborbactam was also bactericidal against strains with meropenem-vaborbactam MICs of 16 mg/liter. In addition, this level of vaborbactam also suppressed the development of resistance observed using phase 1 exposures. In this pharmacodynamic model, exposures similar to 2 g of meropenem in combination with 2 g of vaborbactam administered every 8 h by a 3-h infusion in phase 3 trials produced antibacterial activity and suppressed the development of resistance against carbapenem-resistant KPC-producing strains of Enterobacteriaceae.

Survival Outcomes of Elderly Patients With Glioblastoma Multiforme in Their 75th Year or Older Treated With Adjuvant Therapy.

PURPOSE: To assess the outcomes of the most elderly cohort of patients with a diagnosis of glioblastoma multiforme (GBM) after intensity modulated radiation therapy (IMRT). METHODS AND MATERIALS: The data of patients with GBM who had underwent IMRT from May 2007 to December 2015 were entered into a prospective database. Analysis was performed on the data from patients diagnosed during or after 75 years of age. The primary endpoint was the median survival. Univariate and multivariate analyses were performed with respect to survival for patients aged 74 to 80 versus >80 years, Eastern Cooperative Oncology Group performance status of 0 to 1 versus 2 to 3, extent of resection, a high radiation dose (60 Gy) versus any hypofractionated schedule, MGMT methylation status, planning target volume, and the use of temozolomide (TMZ) versus no TMZ. RESULTS: Of the 108 patients, 35 received best supportive care, 1 received TMZ alone, 40 received RT alone, and 32 received combined RT and TMZ. IMRT was delivered with a hypofractionated technique (40 Gy) in 58 patients or long-course RT (60 Gy) in 11 patients. The median age was 79 years, with 61.6% of patients aged 74 to 80 years and 38.4% aged >80 years. Of the 108 patients, 64 died during the follow-up period, with a median survival of 10 months (95% confidence interval 7.1-11.9), projected 12-month survival rate of 35.6%, and 24-month survival rate of 7.9%. On univariate analysis, the independent predictors of survival included younger age (P=.02), better performance status (P=.014), greater resection extent (P=.002), and TMZ use (P<.001). MGMT methylation status, RT dose, and planning target volume showed no significant differences between the groups. Only chemotherapy use remained statistically significant (P=.035) on multivariate analysis. CONCLUSION: The current data underrepresent elderly patients aged >75 years with GBM. Despite elderly patients having a worse prognosis, the results of the present study suggest the presence of survival benefits with IMRT for selected patients that can be further extended with addition of TMZ. Further study of this cohort and an understanding of the appropriate selection criteria are warranted.