Find open-source science resources

Identifies maximal differential cell populations in flow cytometry data taking into account dependencies between cell populations; flowGraph calculates and plots SpecEnr abundance scores given cell population cell counts.

The creation of effective visualizations is a fundamental component of data analysis. In biomedical research, new challenges are emerging to visualize multi-dimensional data in a 2D space, but current data visualization tools have limited capabilities. To address this problem, we leverage Gestalt principles to improve the design and interpretability of multi-dimensional data in 2D data visualizations, layering aesthetics to display multiple variables. The proposed visualization can be applied to spatially-resolved transcriptomics data, but also broadly to data visualized in 2D space, such as embedding visualizations. We provide this open source R package escheR, which is built off of the state-of-the-art ggplot2 visualization framework and can be seamlessly integrated into genomics toolboxes and workflows.

We propose an Asymmetric Within-Sample Transformation (AWST) to regularize RNA-seq read counts and reduce the effect of noise on the classification of samples. AWST comprises two main steps: standardization and smoothing. These steps transform gene expression data to reduce the noise of the lowly expressed features, which suffer from background effects and low signal-to-noise ratio, and the influence of the highly expressed features, which may be the result of amplification bias and other experimental artifacts.

Huawei's 3D high-resolution global weather forecast model at 0.25° resolution, first AI method to comprehensively outperform traditional NWP across all variables and lead times, integrated into ECMWF operational forecasts (Nature 2023)

This is official model checkpoint for Asclepius-13B (arxiv). This model is the first publicly shareable clinical LLM, trained with synthetic data.

FHIR R4 bundles in JSON format are derived from https://synthea.mitre.org/downloads. Transformation inspired by a kaggle notebook published by Dr Alexander Scarlat, https://www.kaggle.com/code/drscarlat/fhir-starter-parse-healthcare-bundles-into-tables. This is a very limited illustration of some basic parsing and reorganization processes. Additional tooling will be required to move beyond the Synthea data illustrations.

3D Equivariant Diffusion for Target-Aware Molecule Generation (ICLR2023)

Psi4-based reference implementations and Jupyter notebook-based tutorials for foundational quantum chemistry methods.

Physics-informed ML and SciML

This package aims to perform power analysis for the MeRIP-seq study. It calculates FDR, FDC, power, and precision under various study design parameters, including but not limited to sample size, sequencing depth, and testing method. It can also output results into .xlsx files or produce corresponding figures of choice.

Single-cell BERT for gene expression

A preprocessing pipeline for single cell RNA-seq/ATAC-seq data that starts from the fastq files and produces a feature count matrix with associated quality control information. It can process fastq data generated by CEL-seq, MARS-seq, Drop-seq, Chromium 10x and SMART-seq protocols.

PINN research collection

Weather prediction benchmark

Details coming soon

Package that implements the FGGA algorithm. This package provides a hierarchical ensemble method based ob factor graphs for the consistent cross-ontology annotation of protein coding genes. FGGA embodies elements of predicate logic, communication theory, supervised learning and inference in graphical models.

An upgraded causal reasoning tool from Melas et al in R with updated assignments of TFs' weights from PROGENy scores. Optimization parameters can be freely adjusted and multiple solutions can be obtained and aggregated.

Single-cell RNA-sequencing (scRNA-seq) is widely used to explore cellular variation. The analysis of scRNA-seq data often starts from clustering cells into subpopulations. This initial step has a high impact on downstream analyses, and hence it is important to be accurate. However, there have not been unsupervised metric designed for scRNA-seq to evaluate clustering performance. Hence, we propose clustering deviation index (CDI), an unsupervised metric based on the modeling of scRNA-seq UMI counts to evaluate clustering of cells.

The pattern of digestion and protection from DNA nucleases such as DNAse I, micrococcal nuclease, and Tn5 transposase can be used to infer the location of associated proteins. This package contains useful functions to analyze patterns of paired-end sequencing fragment density. VplotR facilitates the generation of V-plots and footprint profiles over single or aggregated genomic loci of interest.

This package can help user to run the m6Aboost model on their own miCLIP2 data. The package includes functions to assign the read counts and get the features to run the m6Aboost model. The miCLIP2 data should be stored in a GRanges object. More details can be found in the vignette.

# Meditron 70B - GGUF - Model creator: EPFL LLM Team - Original model: Meditron 70B

A Vision Transformer (ViT) image classification model. \ Trained on 15M histology patches from PAIP and TCGA. \ Used the MoCo v3 self supervised learning method.

by Sebastian Raschka.

OpenChem is a deep learning toolkit for Computational Chemistry with PyTorch backend.

A list of papers, data sets, and other resources for machine learning for small-molecule drug discovery.

ReUseData is an _R/Bioconductor_ software tool to provide a systematic and versatile approach for standardized and reproducible data management. ReUseData facilitates transformation of shell or other ad hoc scripts for data preprocessing into workflow-based data recipes. Evaluation of data recipes generate curated data files in their generic formats (e.g., VCF, bed). Both recipes and data are cached using database infrastructure for easy data management and reuse. Prebuilt data recipes are available through ReUseData portal ("https://rcwl.org/dataRecipes/") with full annotation and user instructions. Pregenerated data are available through ReUseData cloud bucket that is directly downloadable through "getCloudData()".

This model was finetuned on concatenated pairs of interacting proteins in much the same way as PepMLM. It is meant to generate interaction partners for proteins using the masked language modeling capabilities of ESM-2. The model is not well tested, so use with caution.

cogena is a workflow for co-expressed gene-set enrichment analysis. It aims to discovery smaller scale, but highly correlated cellular events that may be of great biological relevance. A novel pipeline for drug discovery and drug repositioning based on the cogena workflow is proposed. Particularly, candidate drugs can be predicted based on the gene expression of disease-related data, or other similar drugs can be identified based on the gene expression of drug-related data. Moreover, the drug mode of action can be disclosed by the associated pathway analysis. In summary, cogena is a flexible workflow for various gene set enrichment analysis for co-expressed genes, with a focus on pathway/GO analysis and drug repositioning.

Secure text-to-visualization through standardized chart specifications

ProstT5 is a protein language model (pLM) which can translate between protein sequence and structure. !ProstT5 pre-training and inference

## Model Description The "Bird Species Classifier" is a state-of-the-art image classification model designed to identify various bird species from images. It uses the EfficientNet architecture and has been fine-tuned to achieve high accuracy in recognizing a wide range of bird species.

DGL-LifeSci is a [DGL](https://www.dgl.ai/)-based package for various applications in life science with graph neural network.

GEOexplorer is a webserver and R/Bioconductor package and web application that enables users to perform gene expression analysis. The development of GEOexplorer was made possible because of the excellent code provided by GEO2R (https: //www.ncbi.nlm.nih.gov/geo/geo2r/).

Trendy implements segmented (or breakpoint) regression models to estimate breakpoints which represent changes in expression for each feature/gene in high throughput data with ordered conditions.

QuiltNet-B-32 is a CLIP ViT-B/32 vision-language foundation model trained on the Quilt-1M dataset curated from representative histopathology videos. It can perform various vision-language processing (VLP) tasks such as cross-modal retrieval, image classification, and visual question answering.

GNOSIS incorporates a range of R packages enabling users to efficiently explore and visualise clinical and genomic data obtained from cBioPortal. GNOSIS uses an intuitive GUI and multiple tab panels supporting a range of functionalities. These include data upload and initial exploration, data recoding and subsetting, multiple visualisations, survival analysis, statistical analysis and mutation analysis, in addition to facilitating reproducible research.

Question Answering Model for the PathoTHREAT Project

Write-once-read-many table for large datasets.

This package contains infrastructure for benchmarking analysis methods and access to single cell mixture benchmarking data. It provides a framework for organising analysis methods and testing combinations of methods in a pipeline without explicitly laying out each combination. It also provides utilities for sampling and filtering SingleCellExperiment objects, constructing lists of functions with varying parameters, and multithreaded evaluation of analysis methods.

A Chemical Knowledge Graph and Toolkit, writting in IUPAC/SMILES/SMARTS, for common small molecules from diverse communities to aid users in selecting compounds for forcefield parametirization.

First foundation model for weather and climate by Microsoft, Vision Transformer-based architecture trained on heterogeneous datasets (ICML 2023)

Pre-trained weights and exported models for our spine segmentation project. The source code, designed to reproduce our test results and facilitate training and running inference on your own data, is available on GitHub: https://github.com/MMIV-ML/fastMONAI/tree/master/research

MentaLLaMA-chat-7B is part of the MentaLLaMA project, the first open-source large language model (LLM) series for interpretable mental health analysis with instruction-following capability. This model is finetuned based on the Meta LLaMA2-chat-7B foundation model and the full IMHI instruction…

A fuzzy Bruijn graph approach to long noisy reads assembly

CNVfilteR identifies those CNVs that can be discarded by using the single nucleotide variant (SNV) calls that are usually obtained in common NGS pipelines.

Provides Bayesian PCA, Probabilistic PCA, Nipals PCA, Inverse Non-Linear PCA and the conventional SVD PCA. A cluster based method for missing value estimation is included for comparison. BPCA, PPCA and NipalsPCA may be used to perform PCA on incomplete data as well as for accurate missing value estimation. A set of methods for printing and plotting the results is also provided. All PCA methods make use of the same data structure (pcaRes) to provide a common interface to the PCA results. Initiated at the Max-Planck Institute for Molecular Plant Physiology, Golm, Germany.