Bioinformatics Advances

GitHub - CBRC-lab/SiaScoreNet Contribute to CBRC-lab/SiaScoreNet development by creating an account on GitHub.

🧰 Data and sourcecode available here:

Using heterogeneous Siamese neural network architecture, the authors benchmark performance across datasets, evaluate generalization to unseen HLA subtypes, analyze peptide similarity effects, and compare against 14 state-of-the-art predictors, showing strong accuracy, recall, and F1 improvements.

SiaScoreNet is a three-branch pipeline combining ESM-derived embeddings of HLA and peptide sequences with prediction scores from nine IEDB models.

🧪 Now published in Bioinformatics Advances: “SiaScoreNet: A Siamese neural network-based model integrating prediction scores for HLA-peptide interaction prediction.” 

Read the full paper at https://doi.org/10.1093/bioadv/vbaf248

🛠️ Software is available at https://github.com/AnniceNajafi/SurprisalAnalysis. A web-based application with a Graphical User Interface can be found here:

The authors detail the mathematical framework, evaluate normalization and zero-handling approaches, and demonstrate utility across datasets including cancer transcriptomics and T-helper cell differentiation, showing how constraint patterns correspond to coordinated biological processes.

The paper introduces SurprisalAnalysis, an #R package and web application implementing surprisal analysis to decompose #geneexpression into baseline and constraint-driven components.

🧬 Just out in Bioinformatics Advances: “SurprisalAnalysis: An open-source software for information-theoretic analysis of gene expression.” 

Explore the full study at https://doi.org/10.1093/bioadv/vbaf291

🛠️ Software available: https://github.com/Computational-Biology-Aachen/MxlPy 
Documentation can be found here:

The authors evaluate these components across case studies in plant physiology and photosynthesis, demonstrating flexible hybrid modelling and uncertainty quantification.

This work presents MxlPy, a #Python framework that integrates mechanistic modelling with #machinelearning. It provides modules for ODE-based model construction, ensemble simulations, surrogate model integration, NPE workflows, and database-driven parameterisation.

🌱 Explore the latest from Bioinformatics Advances: “MxlPy: Python package for mechanistic learning and hybrid modelling in life science.” 

Full article available at https://doi.org/10.1093/bioadv/vbaf294

uc-ctds (Center for Translational Data Science) Org profile for Center for Translational Data Science on Hugging Face, the AI community building the future.

🧰 Source code is available at: 
🔹https://github.com/uc-cdis/gdc-cohort-copilot
🔹https://huggingface.co/spaces/uc-ctds/GDC-Cohort-Copilot

GDC Cohort LLM weights are available at

They also provide a containerized Gradio app that integrates directly with the GDC interface for refinement and export.

The authors train and compare multiple #LLM architectures, evaluate model performance across real and synthetic datasets, and demonstrate that their GPT-2–based model outperforms GPT-4o for generating accurate case-retrieval cohorts.

The study introduces an open-source tool that uses a custom-trained #largelanguagemodel to translate natural language cohort descriptions into valid GDC cohort filters.

🧪 Now published in Bioinformatics Advances: “GDC cohort copilot: An AI copilot for curating cohorts from the Genomic Data Commons.” 

Read the full paper here: https://doi.org/10.1093/bioadv/vbaf295

GitHub - BioRadOpenSource/ish: Alignment-based filtering CLI tool Alignment-based filtering CLI tool. Contribute to BioRadOpenSource/ish development by creating an account on GitHub.

🛠️ Source code and documentation are available at  https://github.com/BioRadOpenSource/ish, with GPU, scoring matrix, and benchmarking support provided under the Apache-2.0 License.

Benchmarks show that ish matches or exceeds Parasail performance and outperforms agrep and glsearch36 across diverse scenarios, including multi-threaded and GPU runs.

This paper introduces ish, a Unix-style tool for index-free approximate matching using exact local and semi-global alignment. It applies SIMD and GPU-accelerated dynamic programming, supports multiple scoring matrices, and filters sequence records based on normalized alignment thresholds.

🧬Explore the latest from Bioinformatics Advances: “Ish: SIMD and GPU accelerated local and semi-global alignment as a CLI filtering tool.”  

Read the full paper here: https://doi.org/10.1093/bioadv/vbaf292

Author: @ducktapeprogrammer.bsky.social

GitHub - AIGeneRegulation/Sequencing-Data-Manager Contribute to AIGeneRegulation/Sequencing-Data-Manager development by creating an account on GitHub.

💻 SeqManager is open source under the MIT license:

It uses header-aware typing, multi-stage hashing, and conservative regeneration rules to flag true duplicates and safely removable files. Benchmarks across multiple genomics environments show fast scan times, low memory use, and substantial storage savings.

SeqManager is a web-based system built with Flask and React that automates file classification, duplicate detection, and identification of erasable intermediates in sequencing workflows.

🗂️ Explore the latest from Bioinformatics Advances: “SeqManager: A web-based tool for efficient sequencing data storage management and duplicate detection.”

Explore the full study: https://doi.org/10.1093/bioadv/vbaf282

Authors include: @serialchiper.bsky.social

GitHub - OndrejSladky/fmsi: FMSI is a highly memory efficient exact k-mer set index based on masked superstrings and the masked Burrows-Wheeler transform FMSI is a highly memory efficient exact k-mer set index based on masked superstrings and the masked Burrows-Wheeler transform - OndrejSladky/fmsi

🧰 FMSI is developed in C++ and is provided on Github under the MIT licence. Find it here:

It supports membership, dictionary, and streaming queries without requiring (k−1)-overlaps, and benchmarks across genomic, pangenomic, and metagenomic datasets show 2–3x smaller space usage than existing methods with competitive query performance.

FMSI introduces a space-efficient index for arbitrary k-mer sets by combining approximately shortest masked superstrings with a Masked Burrows-Wheeler Transform.

🧮 Just out in Bioinformatics Advances: “FroM Superstring to Indexing: A space-efficient index for unconstrained k-mer sets using the Masked Burrows-Wheeler Transform (MBWT)” 

Full article available: https://doi.org/10.1093/bioadv/vbaf290 

Authors include: @pavelvesely.bsky.social, @brinda.eu

sveva bonomi / PiP2.0 · GitLab GitLab.com

🛠️The AutoDock Vina automation pipeline is freely available for non-commercial use: