deep-learning-atac

Name: deep-learning-atac
Author: GPTomics

1.1kstars185forks

How to install deep-learning-atac?

gptomics/bioskills/deep-learning-atac

$npx -y skills add GPTomics/bioSkills --skill deep-learning-atac

Installs into the current project.

Summary

Sequence-based deep learning for ATAC-seq using chromBPNet, BPNet, scBasset, or Enformer. Use when correcting Tn5 bias with neural networks beyond k-mer models, predicting per-base accessibility profiles, scoring in silico variant effects at GWAS or rare-variant SNPs, discovering

Files · 1

View on GitHub

SKILL.md

1	## Version Compatibility
2
3	Reference examples tested with: chrombpnet 0.1.7+, bpnet-lite 0.6+ (github.com/jmschrei/bpnet-lite), scBasset 0.1.0+ (basenji2 fork), tangermeme 0.1+, tfmodisco-lite 2.2+, DeepLIFT 0.6+, captum 0.7+, tensorflow 2.13+, pytorch 2.1+, kipoi 0.8+.
4
5	Verify before use:
6	- Python: `pip show <package>` then `help(module.function)` to check signatures
7	- CLI: `<tool> --version` then `<tool> --help` to confirm flags
8
9	If code throws unexpected errors, introspect the installed package and adapt rather than retrying. Deep-learning tooling evolves rapidly; method papers post 2023 may have superseded reference implementations.
10
11	# Sequence-Based Deep Learning for ATAC-seq
12
13	"Score the effect of a GWAS SNP on chromatin accessibility" -> Train (or use pre-trained) sequence-to-accessibility CNNs that take 1-5 kb DNA windows and predict per-base Tn5 cleavage profiles. Outputs include: bias-corrected accessibility, single-base mutation effect predictions, and DeepLIFT contribution scores convertible to motifs via TF-MoDISco.
14
15	- CLI: `chrombpnet pipeline --bigwig signal.bw --bigwig-bias bias.bw ...`
16	- Python: `bpnet-lite` for custom architectures; `tangermeme` for fast scoring
17	- Python (single-cell): `scBasset` for per-cell sequence-based predictions
18	- Python (long-context): Enformer pre-trained models via Kipoi
19
20	Sequence models are NOT a replacement for MACS+TOBIAS at every step. They excel at three specific tasks where classical pipelines struggle: (1) Tn5 bias correction in low-complexity sequence contexts, (2) variant effect prediction in non-genic regions, (3) cell-type-specific motif discovery beyond what JASPAR provides.
21
22	## Algorithmic Taxonomy
23
24	\| Tool \| Architecture \| Training \| Output \| Strength \| Fails when \|
25	\|------\|-------------\|----------\|--------\|----------\|------------\|
26	\| chromBPNet (Pampari 2024 bioRxiv) \| Two-track CNN: bias model + accessibility model; bias trained on naked-DNA control or k-mer baseline, accessibility trained on chromatin signal \| Per-cell-type, paired bias track \| Bias-corrected per-base profile + total counts \| Strongest bias correction of the compared tools; established in Kundaje lab pipelines \| Requires GPU, ~24h training per cell type; needs >= 50M reads \|
27	\| BPNet (Avsec 2021 Nat Genet 53:354) \| Original counts + profile dual-head CNN \| TF ChIP-seq or ATAC \| Per-base profile prediction \| Foundational; widely cited; bpnet-lite reimpl maintained \| Less polished than chromBPNet for ATAC; bias correction needs separate model \|
28	\| scBasset (Yuan & Kelley 2022) \| Basenji2-derived CNN, per-cell projection layer \| Single-cell ATAC \| Per-cell sequence-derived peak score \| First sequence model that predicts per-cell accessibility; outperforms chromVAR for cluster discrimination \| Fixed architecture, hard to extend; benchmarks evolving \|
29	\| Enformer (Avsec 2021 Nat Methods 18:1196) \| Long-context Transformer (196 kb input) \| Reference epigenome (DNase + histones + CAGE) \| Per-bin epigenome prediction \| Best for distal regulation modeling; pre-trained available \| Pre-trained models cell-line specific; finetuning on custom data is expensive \|
30	\| Borzoi (Linder 2025 Nat Genet) \| Enformer extension trained on RNA + ATAC \| Multi-tissue paired data \| Sequence -> RNA + chromatin \| Current best benchmark for variant effect on RNA via ATAC linkage \| Newer; benchmarks still emerging \|
31	\| DeepATAC / Basset (legacy) \| Earlier CNN architectures \| -- \| Binary peak prediction \| Historical context; cited in older literature \| Superseded by chromBPNet + Enformer; do not use for new work \|
32	\| tangermeme \| Inference-only fast wrapper \| Use any saved model \| Marginal scoring of variants \| Speeds up variant effect prediction 100x; works with chromBPNet/BPNet outputs \| Inference only; cannot train \|
33
34	Methodology evolves; verify against current Kundaje lab pipelines (chrombpnet GitHub), Greenleaf lab (scBasset), and Avsec / Linder publications before locking pipelines.
35
36	## When Deep Learning Helps vs When Classical Pipelines Suffice
37
38	\| Task \| Classical \| Deep Learning \|
39	\|------\|-----------\|---------------\|
40	\| Peak calling \| MACS3 / Genrich (sufficient) \| chromBPNet (overkill unless variant downstream) \|
41	\| Tn5 bias correction at TF motifs \| TOBIAS ATACorrect (good) \| chromBPNet (better at hard cases: low-complexity flanks, deep TF footprints) \|
42	\| Differential accessibility \| DiffBind / DESeq2 (sufficient) \| -- (no clear DL advantage) \|
43	\| GWAS variant effect prediction at causal SNPs \| Li

People who liked this also viewed

6 picks

Type

deep-learning-atac

1.1kstars185forks

View on GitHub

How to install deep-learning-atac?

gptomics/bioskills/deep-learning-atac

$npx -y skills add GPTomics/bioSkills --skill deep-learning-atac

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	## Version Compatibility
2
3	Reference examples tested with: chrombpnet 0.1.7+, bpnet-lite 0.6+ (github.com/jmschrei/bpnet-lite), scBasset 0.1.0+ (basenji2 fork), tangermeme 0.1+, tfmodisco-lite 2.2+, DeepLIFT 0.6+, captum 0.7+, tensorflow 2.13+, pytorch 2.1+, kipoi 0.8+.
4
5	Verify before use:
6	- Python: `pip show <package>` then `help(module.function)` to check signatures
7	- CLI: `<tool> --version` then `<tool> --help` to confirm flags
8
9	If code throws unexpected errors, introspect the installed package and adapt rather than retrying. Deep-learning tooling evolves rapidly; method papers post 2023 may have superseded reference implementations.
10
11	# Sequence-Based Deep Learning for ATAC-seq
12
13	"Score the effect of a GWAS SNP on chromatin accessibility" -> Train (or use pre-trained) sequence-to-accessibility CNNs that take 1-5 kb DNA windows and predict per-base Tn5 cleavage profiles. Outputs include: bias-corrected accessibility, single-base mutation effect predictions, and DeepLIFT contribution scores convertible to motifs via TF-MoDISco.
14
15	- CLI: `chrombpnet pipeline --bigwig signal.bw --bigwig-bias bias.bw ...`
16	- Python: `bpnet-lite` for custom architectures; `tangermeme` for fast scoring
17	- Python (single-cell): `scBasset` for per-cell sequence-based predictions
18	- Python (long-context): Enformer pre-trained models via Kipoi
19
20	Sequence models are NOT a replacement for MACS+TOBIAS at every step. They excel at three specific tasks where classical pipelines struggle: (1) Tn5 bias correction in low-complexity sequence contexts, (2) variant effect prediction in non-genic regions, (3) cell-type-specific motif discovery beyond what JASPAR provides.
21
22	## Algorithmic Taxonomy
23
24	\| Tool \| Architecture \| Training \| Output \| Strength \| Fails when \|
25	\|------\|-------------\|----------\|--------\|----------\|------------\|
26	\| chromBPNet (Pampari 2024 bioRxiv) \| Two-track CNN: bias model + accessibility model; bias trained on naked-DNA control or k-mer baseline, accessibility trained on chromatin signal \| Per-cell-type, paired bias track \| Bias-corrected per-base profile + total counts \| Strongest bias correction of the compared tools; established in Kundaje lab pipelines \| Requires GPU, ~24h training per cell type; needs >= 50M reads \|
27	\| BPNet (Avsec 2021 Nat Genet 53:354) \| Original counts + profile dual-head CNN \| TF ChIP-seq or ATAC \| Per-base profile prediction \| Foundational; widely cited; bpnet-lite reimpl maintained \| Less polished than chromBPNet for ATAC; bias correction needs separate model \|
28	\| scBasset (Yuan & Kelley 2022) \| Basenji2-derived CNN, per-cell projection layer \| Single-cell ATAC \| Per-cell sequence-derived peak score \| First sequence model that predicts per-cell accessibility; outperforms chromVAR for cluster discrimination \| Fixed architecture, hard to extend; benchmarks evolving \|
29	\| Enformer (Avsec 2021 Nat Methods 18:1196) \| Long-context Transformer (196 kb input) \| Reference epigenome (DNase + histones + CAGE) \| Per-bin epigenome prediction \| Best for distal regulation modeling; pre-trained available \| Pre-trained models cell-line specific; finetuning on custom data is expensive \|
30	\| Borzoi (Linder 2025 Nat Genet) \| Enformer extension trained on RNA + ATAC \| Multi-tissue paired data \| Sequence -> RNA + chromatin \| Current best benchmark for variant effect on RNA via ATAC linkage \| Newer; benchmarks still emerging \|
31	\| DeepATAC / Basset (legacy) \| Earlier CNN architectures \| -- \| Binary peak prediction \| Historical context; cited in older literature \| Superseded by chromBPNet + Enformer; do not use for new work \|
32	\| tangermeme \| Inference-only fast wrapper \| Use any saved model \| Marginal scoring of variants \| Speeds up variant effect prediction 100x; works with chromBPNet/BPNet outputs \| Inference only; cannot train \|
33
34	Methodology evolves; verify against current Kundaje lab pipelines (chrombpnet GitHub), Greenleaf lab (scBasset), and Avsec / Linder publications before locking pipelines.
35
36	## When Deep Learning Helps vs When Classical Pipelines Suffice
37
38	\| Task \| Classical \| Deep Learning \|
39	\|------\|-----------\|---------------\|
40	\| Peak calling \| MACS3 / Genrich (sufficient) \| chromBPNet (overkill unless variant downstream) \|
41	\| Tn5 bias correction at TF motifs \| TOBIAS ATACorrect (good) \| chromBPNet (better at hard cases: low-complexity flanks, deep TF footprints) \|
42	\| Differential accessibility \| DiffBind / DESeq2 (sufficient) \| -- (no clear DL advantage) \|
43	\| GWAS variant effect prediction at causal SNPs \| Li

People who liked this also viewed

6 picks

Type