Biology subjects

Shastry, S.

Publications and source records attributed to Shastry, S..

10 recordsLinked to original sources

bioRxiv2026

A male-specific neural circuit for appetitive thirst in Drosophila

Thirst is the brain representation of the need to restore water balance. While environmental and interoceptive sensory neurons that assess water availability and status are known, the central brain representation of thirst that determines behavior in Drosophila is less well understood. We identified a male-specific appetitive thirst circuit in Drosophila using circuit mapping, connectomics, and behavior measurements. Male-specific NPF-M neurons receive thirsty seeking information via Allatostatin A (AstA) neuropeptide released from the Janu-AstA projection neuron. In turn, NPF-M neurons release the neuropeptides sNPF, NPF, and DH31 to promote thirsty water seeking and the neurotransmitter acetylcholine to inhibit it. NPF-M neurons regulate multiple motivated behaviors, including appetitive and consummatory thirst, food foraging, and courtship, driven by distinct input neurons. NPF-M neurons output to appetitive and aversive dopaminergic, mushroom body ongoing behavior integration center, and navigation circuits. This circuit exhibits similarities to those for hungry food foraging and courtship. NPF-M and Janu-AstA form a network with OviIN hub neurons that are sex-independent thirst neurons. We suggest that NPF-M neurons form a circuit branch that funnels male-specific information to tune motivated behavior.

neuroscience↗

bioRxiv2026

Protein solubility depends on centrifugation: Aiki-Sol, a per-regime predictor for E. coli

MotivationSequence-based predictors of recombinant protein solubility in Escherichia coli have plateaued (NESG independent-test AUC 0.760 [->]~ 0.80 over eight years of protein-language-model variants). The plateau hides a latent confound: the centrifugation regime used to separate the soluble from the insoluble fraction is a hidden variable collapsed into a single binary "soluble" label. The proteins biochemistry does not change between regimes; what changes is which fraction of the lysate is recovered as soluble. Existing predictors treat the regime as label noise rather than a feature, and sequence overlap between training and test partitions masks the resulting failure mode. ResultsWe release the Aiki-Sol Dataset, a tiered E. coli solubility corpus: a ~ 85K stringency-annotated benchmark, an Apache-licensed ~ 147K extension adding binary-only-labelled proteins, and a ~ 229K research-tier pool incorporating non-commercially-licensed sources. On the ~ 85K benchmark, scored on sequence-cluster-disjoint partitions, the strongest published binary comparator falls below chance on the 32,000 x g stratum (AUC 0.491 {+/-} 0.020); a fine-tuned ESM-2 650M backbone with five protocol-matched out-puts lifts pooled AUC by +0.108 (paired-bootstrap CI lower bound +0.090). The gain is curation, not architecture: structure-aware predictors given ESMFold structures do not outperform the sequence-only frame, and capacity scaled to 3B parameters does not exceed the conditioned 650M backbone. The released model, Aiki-Sol, jointly supervises five per-stringency outputs alongside a marginal output for stringency-unknown proteins; on five external cohorts it lifts cohort-mean AUC from 0.69-0.70 to 0.825, with a [≥] +0.10-0.16 lift on the three cohorts at measurably-zero training-pool overlap. Availability and implementationAiki-Sol model weights (Apache 2.0), the 147K-row license-clean training pool of the deployment checkpoint (CC BY 4.0), the cluster-disjoint per-stringency 5-fold partition assignments, per-cohort prediction CSVs, and source code for training, inference, and figure reproduction are available at https://github.com/aikium-public/aiki-sol and archived at Zenodo 10.5281/zenodo.20151817. The research-tier 229K checkpoint is released under CC-BY-NC-ND 4.0 (inheriting the most-restrictive upstream-source tier); its training CSV and the 84,809-protein stringency-annotated bench-mark of [§]2.1 mix non-commercial-tier upstream sources and are not redistributed verbatim. Upstream sources are documented in Data availability and SI [§]S1. The deployment artefact is distributed as a Python package (pip install aikisol) with a predict(seq) entry point. Contactvenkatesh@aikium.com. Supplementary informationSupplementary text, figures, and tables are available at Bioinformatics online.

bioinformatics↗

bioRxiv2026

Aiki-GeNano: Multi-Stage Preference Optimization for Generative Design of Developable Nanobodies

Therapeutic nanobodies must combine target binding with biophysical and chemical properties that determine manufacturability, stability, and clinical viability, collectively termed developability, yet most computational design pipelines still treat developability as a post-hoc filter rather than an integrated training objective. We present Aiki-GeNano, a three-stage language-model alignment pipeline for epitope-conditioned nanobody generation that integrates multiple developability signals directly into training, using only sequence information and previously published predictors. Across 65 target epitopes and relative to the supervised baseline, the combined pipeline raised predicted mean melting temperature by 6.6 {degrees}C, halved isomerization-motif severity, reduced deamidation, N-glycosylation sequons and CDR methionine-oxidation motifs, and preserved predicted humanness and solubility. On a shared 10-target GPCR benchmark, Aiki-GeNano achieved the highest predicted melting temperature and the lowest isomerization severity among five contemporary VHH generators. Starting from ProtGPT2 and a 1.35-million-pair binder dataset generated on an mRNA-display platform, the pipeline applies supervised fine-tuning, Direct Preference Optimization on 522,800 pairs ranked by a composite of selectivity, predicted thermal stability, solubility, and humanness, and Group Reward-Decoupled Policy Optimization against six sequence-based rewards (FR2 hydrophobicity, hydrophobic-patch coverage, chemical-liability motifs, Wilkinson-Harrison expression probability, VHH hallmark residues, scaffold integrity). Generated sequences differ from the nearest training sequence by a mean of 8.1-9.0 amino acids out of 126, and two alternative training trajectories converge to distinct amino-acid-composition strategies with similar liability outcomes but different thermal-stability gains, indicating initialization-dependent convergence of the reward-optimized policy. Predicted humanness was preserved at the level of the camelid VHH scaffold of the training library -- a data-side limitation rather than a methodological one, since the framework was effectively constant across all preference pairs. Applicability to the drug discovery and development pipeline, limitations of predicted-property evaluation, and future work are discussed.

bioinformatics↗

bioRxiv2026

Aiki-XP: leakage-controlled multimodal prediction of within-species relative protein expression at pan-bacterial scale

Generalizable protein-expression prediction can accelerate protein engineering, inform disease mechanisms, and help optimize heterologous recombinant protein production. Protein expression is governed by many interacting parameters that no single omics view captures. We develop Aiki-XP, a multimodal platform integrating four biological scales (genome, operon, coding sequence, protein) plus biophysical features across 492,026 genes from 385 bacterial species. Aiki-XP predicts within-species relative abundance (per-species z-score rank), not absolute copies per cell. Under a leakage-controlled gene-operon split Aiki-XP reaches Spearman{rho} nc = 0.592 versus 0.509 for ESM-C 600M alone, and each tier of a monotone protein[->]operon[->]genome deployment ladder yields a statistically significant gain; a five-recipe rank-average ensemble adds a further +0.016. All recipes were locked before external evaluation; transfer to heterologous, cross-species, and novel-phylum benchmarks demonstrates utility and limits. Ablations and scaling experiments identify operon-scale genomic context, not protein-language-model capacity, as the rate-limiting input at this scale; one foundation model per biological scale suffices, with same-scale stacking adding little.

bioinformatics↗

bioRxiv2025

Molecular prosthetics for CFTR designed for anion selectivity outperform amphotericin B in cultured cystic fibrosis airway epithelia

The ion channel-forming natural product amphotericin B (AmB) can serve as a molecular prosthetic for the cystic fibrosis transmembrane conductance regulator (CFTR) anion channel and thereby restore host defenses in cultured cystic fibrosis (CF) airway epithelia. This is despite the fact that the permeability of AmB-based channels favors cations, and these channels lose their capacity to increase airway surface liquid (ASL) pH in CF airway epithelia at high concentrations. We hypothesize that modifying such channels to favor anion permeability would make them more CFTR-like and thus increase their potential therapeutic effects compared to AmB. Here we show that a synthetic derivative of AmB, AmB-AA, which has an added positively charged appendage and forms ion channels with an improved relative permeability to anions, outperformed AmB in increasing the ASL pH in CF airway epithelia at both low and high concentrations. Further modifications led to another AmB derivative, C2epiAmB-AA, that also minimized cholesterol binding and thus toxicity to cultured CF airway epithelia and was an effective surrogate for CFTR in primary cultured airway epithelia from people with CF.

pharmacology and toxicology↗

bioRxiv2025

Sealable capped nanovials for high-throughput screening of cell growth and function

Millions of modular nanoliter-scale compartments that isolate functionally rich single-cell and cell-to-cell communication data can scale biological discovery for the age of AI. Here, we introduce capped nanovials--suspendable, sealable microscale compartments formed by the docking of hydrogel capping particles into bowl-shaped nanovials--as a versatile system for culturing, analyzing, and sorting single cells and small colonies. This two-particle architecture enables localized confinement of cells and secreted products while maintaining compatibility with standard laboratory workflows such as wash and reagent exchange steps, fluorescence microscopy, and flow cytometry. Crucially, these compartments are formed via simple pipetting and centrifugation steps, making the platform highly democratized. We demonstrate the ability of capped nanovials to compartmentalize single mammalian, bacterial, and yeast cells and support growth into colonies, enabling selection based on proliferation and bioproduction. We further show that capped nanovials enhance single-cell secretion assays by reducing molecular crosstalk and increasing signal-to-noise ratios. Importantly, we demonstrate functional co-culture assays by permitting stable confinement of cell pairs, enabling detection and enrichment of antibody-secreting cells based on the ability of their secreted antibodies to activate co-encapsulated reporter T cells, achieving a signal-to-noise ratio of >30 and up to 100% selection purity. By combining the simplicity of standard lab handling with the resolution and throughput of traditional microfluidic compartmentalization approaches, capped nanovials provide a new class of scalable, accessible test tubes for modern single-cell biology.

bioengineering↗

bioRxiv2025

Adaptive Machine Learning Framework enables Unprecedented Yield and Purity of Adeno-Associated Viral Vectors for Gene Therapy

Adeno-associated viral (AAV) vectors for gene therapy are becoming integral to modern medicine, providing therapeutic options for diseases once deemed incurable. Currently, optimizing viral vector purification is a critical bottleneck in the gene therapy industry, impacting product efficacy and safety as well as accessibility and cost to patients. Traditional optimization methods are resource-intensive and often fail to adjust the purification process parameters to maximize the resulting product yield and quality. To address this challenge, we developed a machine learning framework that leverages Bayesian optimization to systematically refine affinity chromatography parameters (sample load, flow rate, and the formulation of chromatographic media) to improve AAV purification. The efficiency of this closed-loop workflow in iteratively optimizing the vectors yield, purity, and transduction efficiency was demonstrated by purifying clinically-relevant serotypes AAV2, AAV5, and AAV9 from HEK293 cell lysates using the affinity adsorbent AAVidity. We show that three cycles of Bayesian optimization elevated yields from a baseline of 70% to 99%, while reducing host-cell impurities by 230-to-400-fold across all serotypes. The optimized parameters consistently produced vectors with high purity and preserved high transduction activity, essential for therapeutic efficacy and safety, demonstrating serotype versatility - a key challenge in AAV manufacturing. By streamlining parameter optimization and enhancing productivity, our adaptive machine learning framework accelerates process development and reduces costs, advancing the accessibility and clinical translation of AAV-based gene therapies.

molecular biology↗

bioRxiv2021

Single-molecule Mechanical Analysis of Strand Invasion in Human Telomere DNA

Telomeres are essential chromosome end capping structures that safeguard the genome from dangerous DNA processing events. DNA strand invasion occurs during vital transactions at telomeres, including telomere length maintenance by the alternative lengthening of telomeres (ALT) pathway. During telomeric strand invasion, a single stranded guanine-rich (G-rich) DNA invades at a complimentary duplex telomere repeat sequence forming a displacement loop (D-loop) in which the displaced DNA consists of the same G-rich sequence as the invading single stranded DNA. Single stranded G-rich telomeric DNA readily folds into stable, compact, structures called G-quadruplexes (GQ) in vitro, and is anticipated to form within the context of a D-loop; however, evidence supporting this hypothesis is lacking. Here we report a magnetic tweezers assay that permits the controlled formation of telomeric D-loops (TDLs) within uninterrupted duplex human telomere DNA molecules of physiologically relevant lengths. Our results are consistent with a model wherein the displaced single stranded DNA of a TDL folds into a GQ. This study provides new insight into telomere structure and establishes a framework for development of novel therapeutics designed to target GQs at telomeres in cancer cells.

biophysics↗

bioRxiv2021

Directional reorientation of migrating neutrophils is limited by suppression of receptor input signaling at the cell rear through myosin II activity

To migrate efficiently to target locations, cells must integrate receptor inputs while maintaining polarity: a distinct front that leads and a rear that follows. Here we investigate what is necessary to overwrite pre-existing front/rear polarity in neutrophil-like HL60 cells migrating inside straight microfluidic channels. Using subcellular optogenetic receptor activation, we show that receptor inputs can reorient weakly polarized cells, but the rear of strongly polarized cells is refractory to new inputs. Transient stimulation reveals a multi-step repolarization process, confirming that cell rear sensitivity to receptor input is the primary determinant of large-scale directional reversal. We demonstrate that the RhoA/ROCK/myosin II pathway limits the ability of receptor inputs to signal to Cdc42 and reorient migrating neutrophils. We discover that by tuning the phosphorylation of myosin regulatory light chain we can modulate the activity and localization of myosin II and thus the amenability of the cell rear to listen to receptor inputs and respond to directional reprogramming.

cell biology↗

bioRxiv2019

High accuracy DNA sequencing on a small, scalable platform via electrical detection of single base incorporations

High throughput DNA sequencing technologies have undergone tremendous development over the past decade. Although optical detection-based sequencing has constituted the majority of data output, it requires a large capital investment and aggregation of samples to achieve optimal cost per sample. We have developed a novel electronic detection-based platform capable of accurately detecting single base incorporations. The GenapSys technology with its electronic detection modality allows the system to be compact, accessible, and affordable. We demonstrate the performance of the system by sequencing several different microbial genomes with varying GC content. The platform is capable of generating up to 2 Gb of high-quality nucleic acid sequence in a single run. We routinely generate sequence data that exceeds 99% raw accuracy with read lengths of up to 175 bp. Average quality scores remain above Q30 (99.9% raw sequencing accuracy) beyond 150 bp, with more than 85% of total bases at or above Q30. The utility of the platform is highlighted by targeted sequencing of the human genome. We show high concordance of SNP detection on the human NA12878 HapMap cell line with data generated on the Illumina sequencing platform. In addition, we sequenced a targeted panel of cancer-associated genes in a well characterized reference standard. With multiple library preparation approaches on this sample, we were able to identify low frequency mutations at expected allele frequencies.

genomics↗