Yeast protein and RNA abundance

copy-yeast-abundance imports Lawless et al., CoPY (2016) through the same dataset API as human, mouse and dog. It includes calibrated protein and peptide quantities, label-free estimates and RNA from budding yeast grown in glucose-limited chemostats. The strain is EUROSCARF Y11335 BY4742, with the reported genotype retained on each sample.

pip install 'orfdata[data]'
orfdata datasets info copy-yeast-abundance
orfdata datasets download copy-yeast-abundance
orfdata datasets pairs copy-yeast-abundance --basis condition_aggregate

The release contains 69,125 observations in a roughly 14 MB compressed download (168 MB installed). Downloads are explicit and checksum-verified through DataCache. Queries work offline after installation.

Source measurements	Coverage	Interpretation
QconCAT SRM protein summaries	1,903 rows	1,167 quantified; 668 upper bounds; 68 unquantifiable
SRM peptide observations	3,836 rows × four cultures and author median	Protein-specific isotope standards; copies/cell, classification and uncertainty
Aggregate RNA	6,581 labels × FPKM and estimated copies/cell	Counts aggregated before FPKM; copies assume 60,000 total mRNAs/cell
Individual RNA libraries	Four × 7,137 tracking IDs	Native FPKM, 95% confidence interval and Cufflinks status
Label-free estimates	3,166 rows × two methods	Q-Exactive internal-standard Top3 and SAX total-protein scaling remain separate

The 55 joined SRM protein groups retain their combined attribution. Label-free identities also remain groups because full membership is unresolved. Two rows share the label YNL239W; both survive unchanged. Upper bounds have status="below_detection", value=None and a detection_limit; missing and unquantifiable values never become zeros. Author-calibrated SRM estimates have origin="observed"; the RNA copies/cell conversion has origin="derived" with its original FPKM input and calibration assumption.

The QconCAT FASTA contains 93 synthetic calibration standards, not native yeast proteins. Four reported peptides do not match their assigned standard; their measurements retain an explicit validation flag. See source reconciliation #34. Raw transitions and light/heavy ratios are not supplied in the processed workbook. SRM raw data are identified by PASS00717; PXD002694 is the label-free deposit. RNA is from GSE73898. Cufflinks LOWDATA rows retain their original positive/zero estimates and confidence bounds in provenance, while the query value remains not_measured because reliable quantification is unresolved.

Comparing RNA and protein

The paper supports same-cell measurements at the source-cohort level. The individual MS-to-RNA replicate crosswalk is unresolved, so comparisons require explicit condition_aggregate selection. Replicate numbers alone never join samples. Label-free methods have different calibrations and no inferred RNA pair.

from orfdata import DatasetRegistry

dataset = DatasetRegistry().open("copy-yeast-abundance")
match = next(
    m for m in dataset.matched_measurements(allowed_bases=["condition_aggregate"])
    if m.relation.left.identifier == "YML035C"  # AMD1, yeast source context
)
protein = match.right[0]  # 3203.943997 copies/cell; author SRM estimate
rna = next(m for m in match.left if m.unit == "copies/cell")  # ~2.11665
ratio = match.ratio(
    protein, rna, measurement_id="AMD1:protein-per-RNA",
    method="native calibrated protein / estimated RNA",
    normalization={"assumed_total_mrna_per_cell": 60000, "level": "source cohort"},
)
print(ratio.value, ratio.unit, ratio.derivation)

The ratio is conditional on the RNA calibration. It is neither a translation rate nor ribosome occupancy. Group relationships cannot yield an individual protein ratio. All quantities keep their source units and assumptions.

References and other species

Both yeast contexts carry taxon 4932 and pinned source hashes. The historical SGD/Bowtie genome and annotation releases are unspecified; assembly, annotation version and reference strain remain null. These source labels support canonical gene/protein relationships but cannot bind a current PyEnsembl transcript until a reference mapping is validated. Sample BY4742 strain is separate from the unknown reference-genome strain. This dataset does not establish which yeast noncanonical ORFs were searched, expressed or absent.

For cross-species comparisons, supply independently validated, pinned orthology records rather than joining gene names. For example, with an installed mammal RNA dataset and a curated JSON list of OrthologyMapping records:

import json
from orfdata import OrthologyMapping, compare_orthologs

with open("curated-yeast-mammal-orthology.json") as handle:
    mappings = [OrthologyMapping.from_dict(row) for row in json.load(handle)]
for comparison in compare_orthologs(dataset, mammal_rna_dataset, mappings):
    print(comparison.left, comparison.right)  # Separate original measurements

The mapping must name both exact references, its release, source checksum, cardinality and confidence. No expression or protein-existence evidence transfers between species, and different cultures/calibrations remain different experiments. See pairing and orthology.

To rebuild the release from the pinned public workbook, standards, metadata and four RNA files, install orfdata[data,build-data] and run orfdata datasets build copy-yeast-abundance. Python uses DatasetRegistry().build("copy-yeast-abundance"). Local verified inputs can be supplied with repeated --source-path ASSET=PATH or source_paths={...}. The workbook and standards use stable versioned PMC objects; the dynamically generated EuropePMC ZIP is not used as a content identity.