Run any Skill in Manus with one click

txgemma-therapeutics

Stars31

Forks8

UpdatedJune 18, 2026 at 19:20

Operate Google TxGemma prediction and chat models for therapeutic property prediction across small molecules, proteins, nucleic acids, diseases, targets, and cell lines. Use when formatting Therapeutics Data Commons tasks, choosing TxGemma model size or variant, running local or Model Garden inference, fine-tuning on private therapeutic data, or evaluating TxGemma in drug-discovery workflows.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

mdbabumiamssm

mdbabumiamssm/LLMs-Universal-Life-Science-and-Clinical-Skills-

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Biological Scientists, All OtherLife, Physical, and Social Science Occupations·SOC 19-1029

File Explorer

3 files

SKILL.md

readonly

name	txgemma-therapeutics
description	Operate Google TxGemma prediction and chat models for therapeutic property prediction across small molecules, proteins, nucleic acids, diseases, targets, and cell lines. Use when formatting Therapeutics Data Commons tasks, choosing TxGemma model size or variant, running local or Model Garden inference, fine-tuning on private therapeutic data, or evaluating TxGemma in drug-discovery workflows.

TxGemma Therapeutics

Use TxGemma for research prioritization and model development. Predictions and conversational rationales are not experimental measurements or mechanistic proof.

Model Selection

Use a predict checkpoint when raw predictive performance on a supported therapeutic task is the priority.
Use a 9B or 27B chat checkpoint for multi-turn explanation or agentic workflows, accepting that conversational flexibility can reduce predictive performance.
Start with 2B for constrained hardware or screening prototypes, then benchmark 9B and 27B only when the expected gain justifies the compute.

Workflow

Define the endpoint, unit, label space, intended population, and downstream decision before choosing a prompt.
Standardize biomedical inputs: canonicalize SMILES while preserving stereochemistry, verify protein or nucleotide sequence alphabet, and normalize entity identifiers.
Use the official TDC prompt template for supported tasks. Preserve instructions, context, question, answer choices, and any few-shot examples exactly.
Select a split that reflects deployment risk: scaffold or cold-start for novel chemistry, target holdout for new proteins, and temporal split for prospective claims.
Pin the model revision, tokenizer, prompt template, decoding settings, and software environment.
Evaluate against specialist baselines with endpoint-appropriate metrics, calibration, confidence intervals, and subgroup or chemical-space slices.
Fine-tune only after establishing a zero-shot baseline and documenting private-data rights, leakage controls, and an external test set.
Route top candidates to orthogonal computational methods and experimental assays.

Guardrails

Do not infer causal mechanisms from generated explanations.
Do not compare outputs across prompt templates without recalibration.
Do not use random splits alone for claims about novel scaffolds, targets, or future trials.
Detect invalid molecules, unsupported sequence characters, truncated prompts, and answer-format violations.
Review the Health AI Developer Foundations terms before deployment or redistribution.
Require assay confirmation before medicinal-chemistry, toxicology, or clinical decisions.

Output Contract

Return the endpoint definition, model and revision, prompt template, split strategy, preprocessing, metrics, calibration, failure analysis, uncertainty, and recommended experimental validation.

Read references/operations.md for model identifiers, prompt formatting, local inference, and canonical sources.

More from this repository

same repository

medsam2-3d-segmentation

mdbabumiamssm/LLMs-Universal-Life-Science-and-Clinical-Skills-

Operate MedSAM2 for promptable segmentation of 3D medical images and medical videos, including CT lesion propagation, MRI volumes, RECIST-guided prompts, efficient CPU-oriented variants, training, and 3D Slicer integration. Use when generating or validating volumetric masks from sparse prompts or propagating masks through image slices or video frames.

2026-06-1831

monai-medical-imaging

mdbabumiamssm/LLMs-Universal-Life-Science-and-Clinical-Skills-

Build reproducible healthcare imaging pipelines with Project MONAI for DICOM, NIfTI, pathology, and multidimensional imaging tasks including preprocessing, augmentation, training, sliding-window inference, evaluation, model bundles, labeling, and deployment. Use when implementing medical image classification, segmentation, registration, detection, generative, or foundation-model workflows in PyTorch.

2026-06-1831

opencrispr-gene-editors

mdbabumiamssm/LLMs-Universal-Life-Science-and-Clinical-Skills-

Evaluate and operate released Profluent OpenCRISPR gene-editing systems, especially OpenCRISPR-1, for controlled research workflows using its published Cas9-like protein, compatible guide RNA designs, protocols, licensing, specificity testing, and experimental validation. Use when comparing OpenCRISPR-1 with SpCas9, planning nonclinical editing studies, or assessing use in nuclease, nickase, deactivated, base, prime, or epigenome-editing contexts.

2026-06-1831

transcriptformer-cell-embeddings

mdbabumiamssm/LLMs-Universal-Life-Science-and-Clinical-Skills-

Operate CZI TranscriptFormer cross-species generative single-cell models to produce cell embeddings, contextual gene embeddings, likelihoods, zero-shot classifiers, disease-state representations, and regulatory analyses from raw-count AnnData files. Use when selecting TF-Sapiens, TF-Exemplar, or TF-Metazoa, processing in- or out-of-distribution species, or scaling embedding extraction across GPUs.

2026-06-1831

protenix-structure-prediction

mdbabumiamssm/LLMs-Universal-Life-Science-and-Clinical-Skills-

Operate ByteDance Protenix-v2 for open biomolecular structure prediction of proteins, antibodies, nucleic acids, ligands, and complexes using JSON inputs, MSA and template features, constraints, and inference-time sampling. Use when running Protenix locally or through its server, comparing AlphaFold3-style open models, or building reproducible co-folding evaluations.

2026-06-1831

medgemma-health-ai

mdbabumiamssm/LLMs-Universal-Life-Science-and-Clinical-Skills-

Build and evaluate medical text and vision applications with Google MedGemma, including MedGemma 1.5 workflows for CT, MRI, whole-slide pathology, longitudinal chest X-rays, lab reports, and EHR text. Use when prototyping, fine-tuning, deploying, or validating MedGemma-based health AI under clinical data and safety controls.

2026-06-1831

name	txgemma-therapeutics
description	Operate Google TxGemma prediction and chat models for therapeutic property prediction across small molecules, proteins, nucleic acids, diseases, targets, and cell lines. Use when formatting Therapeutics Data Commons tasks, choosing TxGemma model size or variant, running local or Model Garden inference, fine-tuning on private therapeutic data, or evaluating TxGemma in drug-discovery workflows.

TxGemma Therapeutics

Use TxGemma for research prioritization and model development. Predictions and conversational rationales are not experimental measurements or mechanistic proof.

Model Selection

Use a predict checkpoint when raw predictive performance on a supported therapeutic task is the priority.
Use a 9B or 27B chat checkpoint for multi-turn explanation or agentic workflows, accepting that conversational flexibility can reduce predictive performance.
Start with 2B for constrained hardware or screening prototypes, then benchmark 9B and 27B only when the expected gain justifies the compute.

Workflow

Define the endpoint, unit, label space, intended population, and downstream decision before choosing a prompt.
Standardize biomedical inputs: canonicalize SMILES while preserving stereochemistry, verify protein or nucleotide sequence alphabet, and normalize entity identifiers.
Use the official TDC prompt template for supported tasks. Preserve instructions, context, question, answer choices, and any few-shot examples exactly.
Select a split that reflects deployment risk: scaffold or cold-start for novel chemistry, target holdout for new proteins, and temporal split for prospective claims.
Pin the model revision, tokenizer, prompt template, decoding settings, and software environment.
Evaluate against specialist baselines with endpoint-appropriate metrics, calibration, confidence intervals, and subgroup or chemical-space slices.
Fine-tune only after establishing a zero-shot baseline and documenting private-data rights, leakage controls, and an external test set.
Route top candidates to orthogonal computational methods and experimental assays.

Guardrails

Do not infer causal mechanisms from generated explanations.
Do not compare outputs across prompt templates without recalibration.
Do not use random splits alone for claims about novel scaffolds, targets, or future trials.
Detect invalid molecules, unsupported sequence characters, truncated prompts, and answer-format violations.
Review the Health AI Developer Foundations terms before deployment or redistribution.
Require assay confirmation before medicinal-chemistry, toxicology, or clinical decisions.

Output Contract

Return the endpoint definition, model and revision, prompt template, split strategy, preprocessing, metrics, calibration, failure analysis, uncertainty, and recommended experimental validation.

Read references/operations.md for model identifiers, prompt formatting, local inference, and canonical sources.