bio-clip-seq-clip-deep-learning
Predict RBP binding from RNA sequence using deep learning models (RBPNet sequence-to-signal, RNAProt RNN, GraphProt2 GCN with structure, DeepCLIP, DeepRiPe multi-modal CNN) for variant-effect prediction, in silico binding-site discovery, model interpretation, and transfer learning from CLIP and RBNS datasets. Use when computational prediction of RBP binding from sequence is needed, evaluating variant effects on binding without further wet-lab experiments, comparing model performance, or training a custom model on ENCODE eCLIP data.
npx skills add BioTender-max/awesome-bio-agent-skills --skill clip-deep-learning --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Helps train or apply several deep-learning models to predict RNA-binding from sequence, supporting per-base binding profiles (RBPNet), binary site binding probabilities (RNAProt, GraphProt2, DeepCLIP, DeepRiPe), and variant-effect analyses. Includes example commands for training and prediction across multiple tools, and a workflow for variant-effect prediction and custom model training.
How it works
- Enumerates multiple model paradigms and their command patterns:
- RBPNet sequence-to-signal: load pretrained model, predict per-base CL distribution for a sequence.
- RNAProt: train with peaks and GC-matched background, then predict on query sequences.
- GraphProt2: train with structure-aware inputs to produce per-base binding profiles.
- DeepCLIP: train and predict binding probabilities from sequence inputs.
- DeepRiPe: train and predict using a multi-modal CNN on sequence, structure, and region.
- Provides concrete Python and CLI code blocks for training/predicting, including data preparation steps and evaluation notes.
- Delineates failure modes (e.g., data imbalance, background mismatch, leakage, GPU requirements) and fixes, plus variant-effect and genome-wide scoring workflows.
When to use it
Use when you need computational prediction of RBP binding from sequence, evaluation of variant effects on binding without wet-lab work, model comparison across architectures, or training a custom model on CLIP/RBNS data. Also applicable for scenarios requiring per-base binding profiles or transfer learning with foundation models.
What it can touch
- Tools listed: Python-based models (RBPNet, RNAProt, GraphProt2, DeepCLIP, DeepRiPe).
- Mentions specific commands and scripts for training/prediction, including:
- RNAProt train, RNAProt predict
- graphprot2 train
- deepclip, from deepripe import DeepRiPe; model.train; model.predict
- import rbpnet; RBPNet.load_pretrained; model.predict
- Data inputs include peaks.bed, background.bed, genome.fa, query.fa, and FASTA/BED formats; uses bedtools and GC-matching steps.
Caveats
- Notes version compatibility requirements (Python packages like PyTorch, TensorFlow, scikit-learn, Biopython, transformers) and the need to verify API signatures.
- Highlights potential issues: training-data imbalance, background mismatch, data leakage, GPU requirement, window size sensitivity, missing pretrained RBPs for less-studied RBPs, and structure-input dependencies for GraphProt2.
## Version Compatibility Reference examples tested with: RBPNet (Jens & Gagneur 2024 github), RNAProt 0.5+, GraphProt2 (Sauer 2022 github), DeepCLIP 1.0+ (Gronning 2020), DeepRiPe (Krakau lab), pytorch 2.2+, tensorflow 2.15+, scikit-learn 1.4+, biopython 1.83+, transformers 4.40+ (for RNA foundation models). Before using code patterns, verify installed versions match. If versions differ: - Python: `pip show <package>` then `help(module.function)` to check signatures - Frameworks: check pytorch / tensorflow versions; reproducibility depends on framework version If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying. # CLIP-seq Deep Learning **"Predict RBP binding from RNA sequence using deep learning"** -> Train or apply neural networks that learn the sequence (and optionally structure) preference of an RBP from CLIP-seq peaks or single-nucleotide crosslink sites. The output is per-base or per-site binding probability for any input sequence, enabling: (a) variant-effect prediction at heterozygous SNPs; (b) in silico binding-site discovery on transcripts not covered by CLIP; (c) sys
- Version Compatibility
- Models Taxonomy
- Critical Choice: Binary Classification vs Sequence-to-Signal
- Variant-Effect Prediction Workflow
- Training a Custom Model (RNAProt Example)
- RBPNet Sequence-to-Signal Workflow
- Per-Tool Failure Modes
- Training data imbalance
- Background mismatch
- Test on training data leak
- GPU requirement underestimated
- Variant-effect prediction window size
- Pretrained models lack the target RBP
- Structure prediction integration
Step 1: Prepare data bedtools getfasta -fi genome.fa -bed peaks.stringent.bed -s -fo peaks.fa bedtools shuffle -i peaks.stringent.bed -g chrom.sizes -incl expressed.bed -seed 42 > bg.bed bedtools getfasta -fi genome.fa -bed bg.bed -s -fo background.fa Step 2: Train RNAProt RNAProt train \ Step 3: Apply to query sequences RNAProt predict \
What does the bio-clip-seq-clip-deep-learning skill do?
Predict RBP binding from RNA sequence using deep learning models (RBPNet sequence-to-signal, RNAProt RNN, GraphProt2 GCN with structure, DeepCLIP, DeepRiPe multi-modal CNN) for variant-effect prediction, in silico binding-site discovery, model interpretation, and transfer learning from CLIP and RBNS datasets. Use when computational prediction of RBP binding from sequence is needed, evaluating variant effects on binding without further wet-lab experiments, comparing model performance, or training a custom model on ENCODE eCLIP data.
How do I install it?
Run `npx skills add BioTender-max/awesome-bio-agent-skills --skill clip-deep-learning --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From BioTender-max/awesome-bio-agent-skills, a repository with 135 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.