This paper introduces GeneGPT, a novel approach for training LLMs to access and utilize NCBI Web APIs in response to genomics inquiries. The GeneTuring tests are tackled by Codex, which employs in-context learning and an augmented decoding algorithm to detect and execute API calls from the NCBI Web APIs. The GeneTuring benchmark's assessment of GeneGPT's performance across eight tasks yields an average score of 0.83. This demonstrably surpasses comparable models including retrieval-augmented LLMs such as the new Bing (0.44), biomedical LLMs like BioMedLM (0.08) and BioGPT (0.04), as well as GPT-3 (0.16) and ChatGPT (0.12). Our subsequent analyses reveal that (1) API demonstrations exhibit strong cross-task generalizability, surpassing documentations in supporting in-context learning; (2) GeneGPT demonstrates generalization to longer chains of API calls and capably addresses multi-hop questions in GeneHop, a novel dataset; (3) Different types of errors are concentrated in distinct tasks, offering valuable insights for future enhancements.
Ecological competition is a driving force shaping the intricate patterns of species diversity and coexistence. A historically significant method for addressing this query has been the utilization of geometric arguments within the context of Consumer Resource Models (CRMs). As a result, generally applicable principles, including Tilman's $R^*$ and species coexistence cones, have been identified. This work extends the previous arguments by presenting a unique geometrical perspective on species coexistence, specifically using convex polytopes to describe the consumer preference space. We demonstrate the utility of consumer preference geometry in anticipating species coexistence, cataloging stable ecological equilibria, and charting transitions between them. The combined effect of these results establishes a qualitatively new means for comprehending species trait significance in ecosystem construction, in alignment with niche theory.
Transcriptional activity often manifests in punctuated bursts, alternating between periods of high production (ON) and inactivity (OFF). Despite our understanding of transcriptional bursts, the regulatory mechanisms dictating their spatiotemporal control of transcriptional activity are still unclear. Live transcription imaging, using single polymerase precision, is applied to key developmental genes in the fly embryo. buy Atuveciclib Quantifiable single-allele transcription rates and multi-polymerase bursts exhibit shared bursting phenomena among all genes, encompassing both temporal and spatial aspects, and considering cis- and trans-perturbations. The allele's ON-probability is considered the principal factor governing the transcription rate, while changes to the transcription initiation rate are comparatively less impactful. The likelihood of an ON state dictates a particular average ON and OFF duration, while maintaining a consistent characteristic burst duration. A convergence of regulatory processes, as shown by our data, has the primary effect on the ON-probability, thus controlling mRNA synthesis rather than adjusting the ON and OFF times for each mechanism. buy Atuveciclib Subsequently, our results encourage and direct future studies into the mechanisms behind these bursting rules and their influence on transcriptional regulation.
Patient alignment in some proton therapy facilities is accomplished through the use of two orthogonal 2D kV images, acquired from fixed oblique angles, due to the unavailability of in-situ 3D imaging technology. The tumor's visibility in kV radiographs is hampered by the compression of the patient's three-dimensional form onto a two-dimensional plane, particularly when the tumor is positioned behind dense anatomical structures, such as bone. This factor can contribute to considerable mistakes in the patient's setup procedure. Reconstructing a 3D CT image from kV images obtained at the treatment isocenter, within the treatment setup, is a potential solution.
Development of an asymmetric autoencoder-like network incorporated vision transformer building blocks. Data collection involved a single head and neck patient, utilizing 2 orthogonal kV images (resolution: 1024×1024 voxels), 1 3D CT scan with padding (512x512x512 voxels) acquired from the in-room CT-on-rails system pre-kV exposure, and 2 digitally-reconstructed radiographs (DRR) (512×512 voxels) created from the 3D CT. Every 8 voxels, we resampled the kV images, while DRR and CT images were resampled every 4 voxels, creating a 262,144-sample dataset. Each image dimension was 128 voxels in each direction. kV and DRR images were used in tandem during training, forcing the encoder to generate a joint feature map from both datasets. During the testing phase, solely independent kV images were employed. Consecutive sCTs, derived from the model and possessing spatial context, were linked together to construct the full-size synthetic CT (sCT). A determination of synthetic CT (sCT) image quality was made through the application of mean absolute error (MAE) and the per-voxel absolute CT number difference volume histogram (CDVH).
The model demonstrated a speed of 21 seconds and a mean absolute error (MAE) of less than 40HU. Analysis of the CDVH data indicated that less than 5% of voxels displayed a per-voxel absolute CT number variation greater than 185 HU.
We developed a patient-specific vision transformer network that demonstrated both accuracy and efficiency in reconstructing 3D CT images from lower-kilovolt images.
A vision transformer network, tailored to individual patients, was created and demonstrated to be both precise and effective in reconstructing three-dimensional computed tomography (CT) images from kilovolt (kV) images.
Understanding how human brains decipher and handle information is of paramount importance. Brain responses to images, as measured by functional MRI, were examined for selectivity and the presence of inter-individual variations. In our pilot experiment, images projected to attain maximum activation using a group-level encoding model elicited stronger responses than images predicted for average activation; the rise in activation showed a positive relationship with the accuracy of the encoding model. Furthermore, aTLfaces and FBA1 demonstrated stronger activation patterns in response to the highest resolution synthetic images, when compared to the highest resolution natural images. The second experiment indicated a relationship where synthetic images derived using a personalized encoding model provoked more significant responses compared to synthetic images created through group-level or other individuals' models. A repeat experiment corroborated the earlier finding that aTLfaces exhibited a stronger bias for synthetic images than natural images. Data-driven and generative approaches, according to our results, offer a possible pathway for modulating macro-scale brain region responses and examining individual differences and functional specializations of the human visual system.
The disparity between subjects often hinders the generalizability of models in cognitive and computational neuroscience trained on a single individual. The ideal neural conversion system, designed to translate neural signals between individuals, is anticipated to create genuine neural signatures of one subject from another's data, effectively addressing the obstacle of individual differences in cognitive and computational frameworks. A novel EEG converter, termed EEG2EEG, is proposed in this study, inspired by the generative modeling techniques employed in computer vision. Training and testing 72 unique EEG2EEG models, each associated with a pair of subjects from 9, was performed using the THINGS EEG2 dataset. buy Atuveciclib We discovered that EEG2EEG effectively learns how neural representations in EEG signals correlate across different subjects, achieving high levels of conversion precision. Beyond that, the EEG signals created reveal a more apparent and detailed portrayal of visual information in contrast to the data extracted from real-world sources. Employing a novel and state-of-the-art methodology, this framework for converting EEG signals into neural representations offers highly flexible, high-performance mappings between individual brains. This offers critical insight into both neural engineering and cognitive neuroscience.
When a living organism engages with its surroundings, it implicitly places a bet. Given an incomplete comprehension of a random world, the organism must select its next step or immediate course of action, a choice that inherently or explicitly presupposes a model of the world's structure. Better environmental statistics can refine betting strategies, but real-world constraints on gathering this information frequently restrict progress. We posit that optimal inference dictates difficulty in inferring 'complex' models due to bounded information, ultimately causing larger prediction errors. We thus propose a principle of 'playing it safe,' by which, in light of finite information-gathering capabilities, biological systems should exhibit a preference for simpler world models, and thereby, implement less hazardous wagering tactics. Bayesian inference reveals an optimally safe adaptation strategy, directly determined by the prior distribution. We then show that, in the context of stochastic phenotypic switching in bacteria, applying our “playing it safe” principle enhances the fitness (population growth rate) of the bacterial community. We contend that the principle generally applies across problems of adaptation, learning, and evolution, illuminating the environments in which organisms can achieve their maximum potential.
Neocortical neuron spiking activity displays a remarkable degree of fluctuation, regardless of whether the networks are stimulated by identical inputs. The hypothesis that these neural networks operate in the asynchronous state is informed by the neurons' approximately Poissonian firing. The independent firing patterns of neurons in the asynchronous state drastically reduce the possibility of a neuron receiving concurrent synaptic inputs.