Biology subjects

Cramer, S.

Publications and source records attributed to Cramer, S..

6 recordsLinked to original sources

bioRxiv2026

Directed Evolution in Codon Space

Directed evolution is commonly used in protein engineering, where mature molecules are routinely improved through iterative local search of amino acid space. Here, we extend this principle to coding DNA. We developed a language-model-guided framework that iteratively refined industry-optimized coding sequences of clinical-stage therapeutics through synonymous exploration of codon space. Across 23 antibody-based therapeutics, SynCodonLM-guided refinement significantly increased recombinant expression in CHO cells for 17 molecules (74% responder rate), without significant compromise of product-quality or biophysical attributes. Moreover, changes in model likelihood predicted expression gains more effectively than heuristic statistical or mRNA-structure descriptors, despite no explicit expression objective. Codon-level likelihood also tracked temporal progression in influenza A H1N1 sequences, indicating the model captures evolutionary signal. These results show that even production-optimized sequences retain accessible fitness in synonymous codon space, establishing directed evolution as a practical strategy to improve biologic expression, a key manufacturing bottleneck, without altering protein sequence.

bioengineering↗

bioRxiv2025

Advancing Codon Language Modeling with Synonymous Codon Constrained Masking

Codon language models offer a promising framework for modeling protein-coding DNA sequences, yet current approaches often conflate codon usage with amino acid semantics, limiting their ability to capture DNA-level biology. We introduce SynCodonLM, a codon language model that enforces a biologically grounded constraint: masked codons are only predicted from synonymous options, guided by the known protein sequence. This design disentangles codon-level from protein-level semantics, enabling the model to learn nucleotide-specific patterns. The constraint is implemented by masking non-synonymous codons from the prediction space prior to softmax. Unlike existing models, which cluster codons by amino acid identity, SynCodonLM clusters by nucleotide properties, revealing structure aligned with DNA-level biology. Furthermore, SynCodonLM outperforms existing models on 6 of 7 benchmarks sensitive to DNA-level features, including mRNA and protein expression. Our approach advances domain-specific representation learning and opens avenues for sequence design in synthetic biology, as well as deeper insights into diverse bioprocesses.

bioinformatics↗

bioRxiv2022

Spinal cord and brain concentrations of riluzole after oral and intrathecal administration: a potential new treatment route for amyotrophic lateral sclerosis

Riluzole is the only treatment known to improve survival in patients with Amyotrophic Lateral Sclerosis (ALS). However, oral riluzole efficacy is modest at best, further it is known to have large inter-individual variability of serum concentration and clearance, is formulated as an oral drug in a patient population plagued with dysphagia, and has known systemic side-effects like asthenia (limiting patient compliance) and elevated liver enzymes. In this context, we postulated that continuous intrathecal (IT) infusion of low doses of riluzole could provide consistent elevations of the drug spinal cord (SC) concentrations at or above those achieved with oral dosing, without increasing the risk for adverse events associated with systemic drug exposure or off-target side effects in the brain. We developed a formulation of riluzole for IT delivery and conducted our studies in purpose-bred hound dogs. Our non-GLP studies revealed that IT infusion alone was able to increase SC concentrations above those provided by oral administration, without increasing plasma concentrations. We then conducted two GLP studies that combined IT infusion with oral administration at human equivalent dose, to evaluate SC and brain concentrations of riluzole along with assessments of safety and tolerability. In the 6-week study, the highest IT dose (0.2 mg/hr) was well tolerated by the animals and increased SC concentrations above those achieved with oral riluzole alone, without increasing brain concentrations. In the 6-month study, the highest dose tested (0.4 mg/hr) was not tolerated and yielded SC significantly above those achieved in all previous studies. Our data show the feasibility and safety profile of continuous IT riluzole delivery to the spinal cord, without concurrent elevated liver enzymes, and minimal brain concentrations creating another potential therapeutic route of delivery to be used in isolation or in combination with other therapeutics."

neuroscience↗

bioRxiv2022

StandardRat: A multi-center consensus protocol to enhance functional connectivity specificity in the rat brain

Task-free functional connectivity in animal models provides an experimental framework to examine connectivity phenomena under controlled conditions and allows comparison with invasive or terminal procedures. To date, animal acquisitions are performed with varying protocols and analyses that hamper result comparison and integration. We introduce StandardRat, a consensus rat functional MRI acquisition protocol tested across 20 centers. To develop this protocol with optimized acquisition and processing parameters, we initially aggregated 65 functional imaging datasets acquired in rats from 46 centers. We developed a reproducible pipeline for the analysis of rat data acquired with diverse protocols and determined experimental and processing parameters associated with a more robust functional connectivity detection. We show that the standardized protocol enhances biologically plausible functional connectivity patterns, relative to pre-existing acquisitions. The protocol and processing pipeline described here are openly shared with the neuroimaging community to promote interoperability and cooperation towards tackling the most important challenges in neuroscience.

neuroscience↗

bioRxiv2019

An open database of resting-state fMRI in awake rats

Rodent models are essential to translational research in health and disease. Investigation in rodent brain function and organization at the systems level using resting-state functional magnetic resonance imaging (rsfMRI) has become increasingly popular, owing to its high spatial resolution and whole-brain coverage. Due to this rapid progress, shared rodent rsfMRI databases can be of particular interest and importance to the scientific community, as inspired by human neuroscience and psychiatric research that are substantially facilitated by open human neuroimaging datasets. However, such databases in rats are still lacking. In this paper, we share an open rsfMRI database acquired in 90 rats with a well-established awake imaging paradigm that avoids anesthesia interference. Both raw and preprocessed data are made publically available. Procedures in data preprocessing to remove artefacts induced by the scanner, head motion, non-neural physiological noise are described in details. We also showcase inter-regional functional connectivity and functional networks calculated from the database.

neuroscience↗

bioRxiv2018

Biomarker-based patient selection improves stroke rehabilitation trial efficiency

Background and PurposeInter-subject variability complicates trials of novel stroke rehabilitation therapies, particularly in the sub-acute phase after stroke. We tested whether selecting patients using motor evoked potential (MEP) status, a physiological biomarker of motor system function, could improve trial efficiency.\n\nMethodsA retrospective analysis of data from 207 patients (103 women, mean (SD) 70.6 (15.1) years) was used to estimate sample sizes and recruitment rates required to detect a 7-point difference between hypothetical control and treatment groups in upper-limb Fugl-Meyer and Action Research Arm Test scores at 90 days post-stroke. Analyses were carried out for the full sample and for subsets defined by motor evoked potential (MEP) status.\n\nResultsSelecting patients according to MEP status reduced the required sample size by 75% compared to an unselected sample. The estimated time needed to recruit the required sample was also reduced by 72% for patients with MEPs, and was increased by 2-3-fold for patients without MEPs.\n\nConclusionsUsing biomarkers to select patients can improve stroke rehabilitation trial efficiency by reducing the sample size and recruitment time needed to detect a clinically meaningful effect of the tested intervention.

neuroscience↗