K-Dense-AI/scientific-agent-skills / skills/gget
gget
VerifiedQueries 20+ bioinformatics resources through CLI/Python. Supports quick lookups of gene info, BLAST/BLAT, viral sequence downloads, PDB/mmCIF structures, G2P residue annotations, enrichment analysis, OpenTargets, COSMIC, CELLxGENE, and 8cube mouse specificity/expression data. Best for interactive exploration and simple queries. For batch processing or advanced BLAST use biopython; for multi-database Python workflows use bioservices.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 11.0 KB |
| references/common_workflows.md | markdown | 2.5 KB |
| references/database_info.md | markdown | 13.3 KB |
| references/module_catalog.md | markdown | 26.0 KB |
| references/module_reference.md | markdown | 23.8 KB |
| references/workflows.md | markdown | 9.6 KB |
| scripts/batch_sequence_analysis.pyexecutable | script | 6.6 KB |
| scripts/enrichment_pipeline.pyexecutable | script | 7.7 KB |
| scripts/gene_analysis.pyexecutable | script | 7.4 KB |
references/workflows.md
View on GitHub# gget workflow examples
These are **illustrative integration recipes**, checked against gget 0.30.8
signatures and response contracts, not fully executed scientific analyses.
Remote results, optional dependencies, licenses, and input files vary. The
[database contracts](database_info.md) record public verification and failures.
Use the bundled scripts for tested export/control-flow behavior.
## Gene discovery to annotations and sequences
`search` matches descriptions and synonyms. A search for `TP53` can return other
genes before TP53; never use the first search hit as an exact identifier mapping.
```python
import gget
from pathlib import Path
symbol = "TP53"
hits = gget.search(symbol, species="homo_sapiens")
if hits is None:
raise RuntimeError("Search failed")
exact = hits.loc[hits["gene_name"].eq(symbol)].drop_duplicates("ensembl_id")
if len(exact) != 1:
raise ValueError("Resolve missing or ambiguous symbol before proceeding")
gene_id = exact.iloc[0]["ensembl_id"]
info = gget.info(gene_id, pdb=True)
if info is None or info.empty:
raise RuntimeError("No annotation returned; inspect service errors")
# Query IDs live in the index. Annotation fields can be missing or multi-valued.
info.to_csv("tp53_info.csv", index_label="query_ensembl_id")
print(info[["primary_gene_name", "ensembl_gene_name", "ensembl_description"]])
protein_fasta = gget.seq(gene_id, translate=True)
if not protein_fasta:
raise RuntimeError("No protein sequence returned")
Path("tp53_protein.fasta").write_text("\n".join(protein_fasta) + "\n")
```
The released Ensembl HTTP adapter failed in this review. Do not treat that as
absence of the gene; resolve transport before running this full recipe.
Nucleotide `gget.seq(gene_id)` is genomic sequence, not an automatically selected CDS.
## Comparative sequences and structures
Choose orthologs from `gget.bgee(..., type="orthologs")`; verify species and
one-to-many orthology before comparing. Save flat FASTA records, not nested
lists of the output from `seq`.
```python
import gget
from pathlib import Path
orthologs = gget.bgee("ENSG00000169174", type="orthologs") # human PCSK9
print(orthologs[["gene_id", "gene_name", "species"]])
# Select a confirmed mouse ortholog from the response, rather than guessing an ID.
mouse = orthologs.loc[orthologs["species"].eq("musculus")]
if len(mouse) != 1:
raise ValueError("Choose the intended mouse ortholog explicitly")
human_lines = gget.seq("ENSG00000169174", translate=True)
mouse_lines = gget.seq(mouse.iloc[0]["gene_id"], translate=True)
if not human_lines or not mouse_lines:
raise RuntimeError("Sequence retrieval failed")
Path("orthologs.fasta").write_text("\n".join(human_lines + mouse_lines) + "\n")
gget.muscle("orthologs.fasta", out="orthologs.afa") # returns None
# G2P connects a specified protein to existing isoforms and structures.
structure_map = gget.g2p("TP53", uniprot_id="P04637", resource="map")
if structure_map is not None:
print(structure_map[["UniProt Isoform", "PDB Ids List"]])
# An explicitly chosen structure, after checking chain and residue coverage:
structure = gget.pdb("1TUP", resource="mmcif")
if structure is not None:
Path("1TUP.cif").write_text(structure)
```
Do not extract a PDB accession by splitting free-text BLAST descriptions.
Check chain sequence, residue numbering, coverage, ligands, experimental method,
and structure quality. Predicted structure confidence is not experimental
validation. The gget AlphaFold wrapper is deprecated and is not part of this recipe.
## Cancer and drug associations
```python
import gget
gene_id = "ENSG00000146648" # EGFR, human
associations = gget.opentargets(gene_id, resource="diseases", limit=10)
print(associations[["disease.id", "disease.name", "score"]])
drugs = gget.opentargets(gene_id, resource="drugs", limit=10)
if not drugs.empty:
print(drugs[["drug.name", "drug.drugType", "drug.maximumClinicalStage"]])
tractability = gget.opentargets(gene_id, resource="tractability")
print(tractability)
expression = gget.opentargets(gene_id, resource="expression", limit=100)
if not expression.empty:
print(expression[["median", "unit", "datasourceId", "datatypeId"]])
interactions = gget.opentargets(gene_id, resource="interactions", limit=10)
print(interactions.reindex(columns=["targetB.id", "targetB.approvedSymbol", "score"]))
```
These are bounded exploratory slices. gget does not traverse Open Targets pages;
local filters cannot retrieve records outside the fetched page. A filter applied
after `limit=10` searches only those ten rows. The disease score does not validate
a target, and an associated drug is not a treatment recommendation. Do not label
candidate targets “validated” from these queries alone.
cBioPortal and COSMIC are separate input routes:
```bash
# Optional dependencies; then discover and select appropriate studies.
gget setup cbio
gget cbio search breast
# Illustrative: downloads study data and writes a heatmap.
gget cbio plot -s msk_impact_2017 -g EGFR TP53 -st tissue -vt mutation_occurrences
# Requires a previously downloaded, licensed COSMIC TSV and matching project.
gget cosmic EGFR --cosmic_project cancer --cosmic_tsv_path cosmic_data.tsv -l 10
```
Record study/sample coverage, assembly, mutation definitions, and denominators.
A missing file or adapter failure is not a zero-mutation result.
## Expression and enrichment
```python
import gget
bulk = gget.archs4("TP53", which="tissue", species="human")
print(bulk[["id", "median"]].head())
correlated = gget.archs4("TP53", which="correlation", gene_count=20)
print(correlated[["gene_symbol", "pearson_correlation"]])
```
For a differential-expression list, supply the genes actually tested as the
background; record mapped/unmapped counts and use an organism-appropriate full
library identifier. The following uploads the lists to Enrichr and is illustrative:
```python
import gget
from pathlib import Path
selected = Path("selected_gene_symbols.txt").read_text().splitlines()
tested = Path("tested_gene_symbols.txt").read_text().splitlines()
if not set(selected).issubset(tested):
raise ValueError("Selected genes must belong to the tested universe")
result = gget.enrichr(selected, database="GO_Biological_Process_2021",
background_list=tested, species="human", plot=False)
if result is None:
raise RuntimeError("Enrichment failed")
result.to_csv("enrichment.csv", index=False)
significant = result.loc[result["adj_p_val"] < 0.05]
print(significant[["path_name", "adj_p_val", "overlapping_genes"]])
```
A coexpression-selected list needs a background from the genes eligible for that
selection. Enrichr returns rows irrespective of your significance threshold;
empty/error responses do not prove absence of enrichment. For fly/yeast/worm/fish,
use a species-specific full library and omit custom backgrounds.
## Scoped single-cell exploration
Install `gget[cellxgene]==0.30.8` in a compatible Python 3.12/3.13 environment.
Choose a retained Census snapshot and an explicit dataset ID first; placeholders
below deliberately require input selection. This example is not a full scRNA-seq
analysis or a tested large download.
```python
import gget
census_version = "2025-11-08" # Example LTS; confirm availability before querying.
dataset_id = "REPLACE_WITH_SELECTED_DATASET_UUID"
metadata = gget.cellxgene(dataset_id=dataset_id, species="homo_sapiens",
census_version=census_version, meta_only=True,
column_names=["dataset_id", "donor_id", "cell_type", "is_primary_data"])
print(f"Selected cells: {len(metadata)}") # gene filtering does not affect this table
adata = gget.cellxgene(gene=["ACE2", "TMPRSS2"], dataset_id=dataset_id,
species="homo_sapiens", census_version=census_version)
print(adata.shape)
print(adata.var[["feature_id", "feature_name"]])
# var_names may be Ensembl IDs, not symbols. Match feature_name explicitly.
for symbol in ["ACE2", "TMPRSS2"]:
selected = adata[:, adata.var["feature_name"].eq(symbol).to_numpy()]
if selected.n_vars == 1 and selected.n_obs:
print(symbol, "raw nonzero fraction", float((selected.X > 0).sum() / selected.n_obs))
```
Do not filter cells at `min_genes=200` after retrieving two/four genes: that
removes every cell. Do not normalize a selected marker panel as if it were a
whole-transcriptome library. For QC/normalization, retrieve the required full
feature universe or valid precomputed whole-cell metrics, and account for assay,
dataset, donor, primary-data duplication, and batch before comparing expression.
## Reference-file preparation
```bash
# Historical pinned example; verify the assembly and release suit the study.
gget ref homo_sapiens -r 110 -w gtf,cdna -d -od reference -o reference_manifest.json
# Python ref returns links/metadata; downloading is a CLI capability.
```
Use a dedicated output directory, record checksums, and build downstream indexes
with the downloaded filenames actually present. Do not mix the pinned release
with unrecorded latest annotations. `search` accepts a release, whereas `info`
and `seq` query current REST data.
## Applying a coding variant
This local synthetic example was executed with 0.30.8:
```python
import gget
cds = "ATCGCTAAGCT"
assert cds[3] == "G" # HGVS c.4 is 1-based
mutated = gget.mutate(cds, "c.4G>T", verbose=False)
assert mutated == ["ATCTCTAAGCT"]
```
For real variants, first obtain the exact transcript-version CDS, confirm the
reference base and reading frame, then apply the variant. Genomic gene sequences,
UTRs, introns, negative-strand coordinates, and different isoforms cannot be used
interchangeably. A sequence edit or structure comparison alone is not evidence of
pathogenicity, frequency, or functional effect.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| deprecation-and-migrationaddyosmani/agent-skillsManages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when migrating a database schema in production, such as renaming or dropping a column without downtime (expand/contract). Use when deciding whether to maintain or sunset existing code. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| performance-optimizationaddyosmani/agent-skillsOptimizes application performance across frontend, backend, queries, and databases. Use when performance requirements exist, when you suspect performance regressions, when Core Web Vitals or load times need improvement, when N+1 query patterns need fixing, or when profiling reveals bottlenecks. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |