Update SDRF annotation for PXD006675 (region- and cell-type-resolved human heart atlas)

[enriquea]

Update SDRF annotation for PXD006675 (Doll et al., human heart atlas)

PXD006675 — Region and
cell-type resolved quantitative proteomic map of the human heart, Doll, Dreßen, Geyer,
… Krane, Mann, Nat Commun 2017, doi:10.1038/s41467-017-01747-2
(PMC5684139, open access).

Same 594 rows and the same 594 raw files as the current annotation — no runs added or
dropped. The changes are to the sample metadata.

Why this update

The dataset's defining axis is anatomical region, but every row currently reads
characteristics[organism part] = heart. The region was present only as a bare code in
characteristics[phenotype], and the atrial-fibrillation arm was attached to the wrong
runs.

	before	after
distinct organism part values	1	20
rows with an ontology-coded organism part	0	518
rows with an ontology-coded disease	0	473
rows carrying a donor age	94	473
factor value[…] columns	0	2

Corrections

1. Region resolved (441 atlas runs). The 16 regions are recoverable from the run
names, where the trailing digit is the donor (3, 4, 8):

LA LV RA RV · Ao PA Vcavainf PVe · TV MV PV AV · SepA SepV · LH CP

Each is now a UBERON term in organism part, e.g. NT=interventricular septum;AC=UBERON:0002094. Structure checks out: 16 regions × 3 donors × ~9 fractions =
441 runs.

2. The atrial-fibrillation arm was on the wrong runs. Supplementary Table 3 names
exactly three AFib patients — A3689 (F, 76, persistent), A2719 (M, 78, permanent),
A2725 (F, 74, persistent) — and Fig. 6a states their LA tissue was run as technical
triplicates. The deposit contains exactly three runs for each of those IDs.

The current file instead labels nine AF_Easy12_14_DMSO_A#### runs as atrial
fibrillation, and labels the three real patients trauma victims. Those AF_… runs
carry donor IDs that appear nowhere in Supplementary Table 3; AF in this deposit is
adipose fibroblast (Methods: AFs isolated from subcutaneous fat, frozen in DMSO),
which collides with the paper's use of "AFib" for the disease. They are now annotated as
AF cell cultures with disease = not available, and the AFib label moved to A3689 /
A2719 / A2725 with their published age, sex and AF type.

3. Six individuals that do not exist have been removed. Each AFib patient's technical
replicates had been given incremented IDs — A3689 → A3690, A3691, A2719 → A2720, A2721, A2725 → A2726, A2727. Those are the same three patients, now annotated as
comment[technical replicate] 1–3.

4. disease = "trauma victims" → normal (PATO:0000461). Supplementary Table 1
records heart morphology "no abnormality detected" for all three donors, and the Methods
state the hearts "did not present any relevant injury or signs of cardiac malfunction and
were therefore defined as healthy". The autopsy provenance is preserved in
characteristics[phenotype] instead of being written into the disease column.

5. Donor demographics propagated. Run-name donor 3/4/8 → Supplementary Table 1
subjects 1/2/3 (male 22 y, male 21 y, male 47 y). The existing file already carried these
ages on 27 runs each, consistently; they now apply to all runs of each donor.

6. Cell types resolved to CL with their true tissue of origin, which is not the heart
for two of them: EC → cardiac endothelial cell (CL:0010008, atrium), CF →
fibroblast of cardiac tissue (CL:0002548, atrium), AF → fibroblast (CL:0000057,
subcutaneous adipose tissue), SMC → smooth muscle cell of the internal thoracic artery (CL:0002593, internal thoracic artery).

7. Technical metadata completed from the Methods. Added Lys-C alongside trypsin
(digestion was Lys-C + trypsin, 1:50 overnight then 1:100 for 3 h), plus dissociation
method, collision energy (27% NCE) and high-pH reversed-phase fractionation.
Acetyl changed from PP=Any N-term to PP=Protein N-term, matching MaxQuant's own
summary.txt in the deposit (Oxidation (M);Acetyl (Protein N-term)). Carbamidomethyl,
Oxidation and the tolerances were already correct and are unchanged.

8. Fraction identifiers now come from the submitter. ExperimentalDesignHeartAtlas.txt
inside the deposit's search.zip gives the per-run fraction; MaxQuant's 100 (meaning
unfractionated) is mapped to fraction identifier 1. This covers 564 of 594 runs; the rest
keep the fraction parsed from the run name.

LH and CP — the one inference in this PR

These are the study's two coronary arteries (the paper lists LCA and RCA among its 16
regions, and after mapping the other 14 tokens these are the only ones left), but the
strings LH and CP appear nowhere in the paper or its supplementary material.
They are annotated LH = left coronary artery (UBERON:0001626) and CP = right coronary artery (UBERON:0001625) on two independent grounds that agree:

1. Column ordering. Five supplementary data sheets and the Supplementary Figure 8
   x-axis all list the regions in the same non-alphabetical order
   Ao AV RCA LA LCA LV MV PA PV PVe RA RV SepA SepV TV IVC. That is exactly the
   case-insensitive alphabetical sort of the internal run tokens
   {Ao AV CP LA LH LV MV PA PV PVe RA RV SepA SepV TV Vcavainf} under LH→LCA, CP→RCA.
   The opposite assignment reproduces none of those orderings.
2. Quantitative correlation. proteinGroups.txt inside the deposit's search.zip
   carries per-donor intensities under LH3/LH4/LH8 and CP3/CP4/CP8. Correlating
   per-region medians against the published LCA/RCA copy-number columns of Supplementary
   Data 3 gives the same assignment, with the other 14 regions as a positive control.

This is the main thing to check. If a reviewer with knowledge of the original sample
naming disagrees, the fix is two lines, and the conservative fallback is the unsided
parent coronary artery (UBERON:0001621) for both.

Other things a reviewer may want to weigh in on

• The 99 pilot / method-development runs (singleShot, BoxCar, urea, test,
  misslabeled, …) are annotated with the region where the run name states one and
  not available otherwise, with phenotype = method-development / pilot run. An
  alternative would be to drop them from the SDRF.
• AF_Easy12_14_DMSO_A#### is read as adipose-fibroblast cultures from six cell donors.
  The reasoning is above; disease is left not available rather than asserting a state.

Validation

parse_sdrf validate-sdrf --sdrf_file datasets/PXD006675/PXD006675.sdrf.tsv --use_ols_cache_only
# Everything seems to be fine. Well done.

• Passes with sdrf-pipelines from GitHub main.
• Every comment[data file] value is unchanged from the current annotation and matches
  the PRIDE deposit; 594 rows, 594 unique files, no duplicates.
• No (source name, fraction identifier, technical replicate, label) collisions.
• Every ontology accession resolved live against OLS4 and re-checked by an independent
  pass; all CL/UBERON terms confirmed non-obsolete via the OLS term endpoint.

Annotated with sdrf-skills.

[qodo-code-review]

Qodo reviews are paused for this user.

Troubleshooting steps vary by plan Learn more →

On a Teams plan?
Reviews resume once this user has a paid seat and their Git account is linked in Qodo.
Link Git account →

Using GitHub Enterprise Server, GitLab Self-Managed, or Bitbucket Data Center?
These require an Enterprise plan - Contact us
Contact us →

[coderabbitai]

Important

Review skipped

Review was skipped due to path filters

⛔ Files ignored due to path filters (1)

• datasets/PXD006675/PXD006675.sdrf.tsv is excluded by !**/*.tsv

CodeRabbit blocks several paths by default. You can override this behavior by explicitly including those paths in the path filters. For example, including **/dist/** will override the default block on the dist directory, by removing the pattern from both the lists.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 10854f00-540b-4cad-b429-c704c2a93263

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[Copilot]

Copilot wasn't able to review any files in this pull request.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[ypriverol]

Some comments @enriquea:

Normally; we use the label style for sample metadata instead of the full term; like:

NT=mitral valve;AC=UBERON:0002135 -> mitral valve

For the samples we use PX{}-Sample{index}

instead of Donor8 -> PXD006675-Sample-{index}

For the individual, we do use another notation, like accession, which enable to know samples that comes from the same patient/individual.

[enriquea]

All three applied here and to #61 and #63, since the same conventions were wrong across all of them.

1. Bare labels. characteristics[*] now carry the label only (NT=mitral valve;AC=UBERON:0002135 → mitral valve). I scoped this to characteristics[*] and left comment[*] as NT=..;AC=.., based on what main actually does — in 220 sampled datasets, comment[instrument] is NT=..;AC=.. in 209 and comment[cleavage agent details] in 212, while the characteristics[*] columns are predominantly bare labels. Say the word if you want comment[*] converted too.

2. Sample accessions. source name is now PXD006675-Sample-<n> (160 samples here), assigned per sample so fractions and technical replicates of one sample keep sharing an identifier. #63 already used this; #61 and #62 have been converted.

3. Individual. This is the one I had to interpret, so flagging it explicitly: characteristics[individual] is corpus-wide free-form today (139 datasets, none accession-style), so there was no existing pattern to copy. I used <ACCESSION>-Individual-<n>, numbered per distinct individual, which keeps same-donor samples linkable and parallels the sample accessions. If you have a different notation in mind, tell me the shape and I'll re-run it — it's a scripted rename across all three PRs.

One consequence worth surfacing: for this dataset the individual column held real donor codes tied to Supplementary Table 3 of the paper, so replacing them loses that link. Recording the mapping here to keep it traceable:

Individual	Original donor code
PXD006675-Individual-1	subject 1
PXD006675-Individual-2	subject 2
PXD006675-Individual-3	subject 3
PXD006675-Individual-4	A2719
PXD006675-Individual-5	A2725
PXD006675-Individual-6	A3689
PXD006675-Individual-7	A21
PXD006675-Individual-8	A2684
PXD006675-Individual-9	A2771
PXD006675-Individual-10	A3602
PXD006675-Individual-11	A3762
PXD006675-Individual-12	A3780

Individuals 4–6 are the atrial fibrillation patients; 1–3 are the healthy multi-region donors. If you'd rather keep the published codes visible, an alternative is PXD006675-Individual-A3689 — that satisfies the accession prefix while staying greppable against the paper.

All three PRs revalidate clean after the change (19/19, 594/594 rows, 14/14) with no coordinate collisions introduced.

Two things unrelated to your comment that need a maintainer decision:

• CodeRabbit reviews nothing in this repo. There's no .coderabbit.yaml, so it runs on defaults, and the default filters exclude !**/*.tsv — every PR here gets a green "Review completed" having read zero files. A one-line config adding **/*.tsv would fix it.
• The judgment calls I flagged in the PR bodies still need a domain opinion, particularly the LH/CP → left/right coronary artery inference in this PR.