Added the two datasets from the surf publication (#195)

[bdeadman]

Borylation and minisci datasets from the SURF publication (ChemRxiv, 2024, 10.26434/chemrxiv-2023-nfq7h-v2 D O I: 10.26434/chemrxiv-2023-nfq7h-v2 [opens in a new tab]). These are reactions which have been collected from the literature and summarised in SURF format by @alexarnimueller.

The Jupyter Notebook used to convert the datasets is located at bdeadman/surf/surf2ord_troubleshooting.ipynb. The surf2ord.py script has been modified to output data into the latest ord-schema version and preferred style.

Notes:

• Provenance data was not found in the Minisci dataset so this is assumed to also be @alexarnimueller
• In the borylation dataset several rows had catalyst_1 only defined by the CAS number. In all but 1 of these I have found a SMILES string to approximate the catalyst.
• surf2ord now assigns each reagent/catalyst/reactant/solvent to a separate input instead of collecting together them by role. Inputs with multiple components would be used when they are known to be added as a solution or mixture.
• rxn_type has been converted to REACTION_TYPE instead of NAME (makes it compatible with ord-schema >0.3.38)
• cas numbers have been converted to the CAS_NUMBER type instead of NAME
• Isolated analysis type in SURF has been defined as WEIGHT analysis type in ORD.
• datast_name and dataset_description options added to surf2ord function. This ensures the output dataset passes validations in ord-schema >0.3.86. Placeholder text is included by default so the text can be edited afterwards.
• Fixed the code so it no longer multiples fractional yields by 100 * 100.

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/borylation_ord.pbtxt	0	0	0
data/minisci_ord.pbtxt	0	0	0
	0	0	0

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/c1/ord_dataset-c1e2bd4243aa49448b1e61636463c7cf.pb.gz	130	0	0
	1241	0	0

[bdeadman]

@connorcoley @skearnes I was able to pull this through into open-reaction-database/ord-data:#195 without getting approval. It should probably be checked at this stage before it is approved to go into main.

[skearnes]

@connorcoley @skearnes I was able to pull this through into open-reaction-database/ord-data:#195 without getting approval. It should probably be checked at this stage before it is approved to go into main.

Yes, that's expected; we don't protect any branches except for main by requiring approvals.

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/c1/ord_dataset-c1e2bd4243aa49448b1e61636463c7cf.pb.gz	130	0	0
	1241	0	0

[skearnes]

@qai222 do you have time to take a look at these for correctness?

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/c1/ord_dataset-c1e2bd4243aa49448b1e61636463c7cf.pb.gz	130	0	0
	1241	0	0

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/c1/ord_dataset-c1e2bd4243aa49448b1e61636463c7cf.pb.gz	130	0	0
	1241	0	0

[qai222]

I reviewed the pbtxt files from #195, specifically borylation_ord.pbtxt and minisci_ord.pbtxt.

1. not sure why the analysis type for yield is isolated

       analyses {
     key: "product_1_isolated"
     value {
       type: CUSTOM
       details: "isolated"
     }
   

2. some reactions have yield higher than 100

         identifiers {
       type: CUSTOM
       details: "rxn_id from SURF table"
       value: "lit_pub_bo_9"
       }
       ...
         measurements {
       analysis_key: "product_1_GC"
       type: YIELD
       percentage {
         value: 137.0
       }
     }
   

[bdeadman]

For item 2, this is how it is recorded in the SURF table file. Since it is a GC yield I suspect it is just a calibration error. While not ideal, I think we need to report it as it is written. There are already >100% yields in the ORD from the USPTO data.

This particular reaction has come from this paper: https://pubs.acs.org/doi/10.1021/acscatal.0c00152. Unfortunately the rxn does not appear in the SI, and I don't have access to the paper.

[qai222]

For item 2, this is how it is recorded in the SURF table file. Since it is a GC yield I suspect it is just a calibration error. While not ideal, I think we need to report it as it is written. There are already >100% yields in the ORD from the USPTO data.

This particular reaction has come from this paper: https://pubs.acs.org/doi/10.1021/acscatal.0c00152. Unfortunately the rxn does not appear in the SI, and I don't have access to the paper.

Yeah the paper says "Yields were determined by gas chromatography and are based on moles of B 2 pin 2." in table 2 caption. I agree we should report as it is.

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/c1/ord_dataset-c1e2bd4243aa49448b1e61636463c7cf.pbtxt	0	0	0
	1111	0	0

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/c1/ord_dataset-c1e2bd4243aa49448b1e61636463c7cf.pb.gz	130	0	0
data/fe/ord_dataset-feaf1b793c6d408aaec1cac7cc3ceadc.pb.gz	130	0	0
	1371	0	0

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/c1/ord_dataset-c1e2bd4243aa49448b1e61636463c7cf.pb.gz	130	0	0
	1241	0	0

[bdeadman]

@qai222 I fixed the surf2ord program to correctly label "isolated" yield types as the WEIGHT measurement type enum in ORD. The problem was some rogue capitalization in their code.

As discussed above, point 2 (the >100% yields) are left as they were reported.

I made a bit of a mess when replacing the minisci dataset file, but it looks like it has been resolved now, and the PR only shows the 2 datasets and the reaction count is correct. I downloaded them to confirm they are the correct versions.

[bdeadman]

@skearnes these two have been peer-reviewed by @qai222, and now need the PR to be approved by a reviewer with write access.

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/fe/ord_dataset-feaf1b793c6d408aaec1cac7cc3ceadc.pb.gz	130	0	0
	1241	0	0

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/fe/ord_dataset-feaf1b793c6d408aaec1cac7cc3ceadc.pb.gz	130	0	0
	1241	0	0

[github-actions]

Change summary:

Filename	Added	Removed	Changed
data/99/ord_dataset-99c23cf435dc42f1af884053bc8b11c7.pb.gz	1111	0	0
data/fe/ord_dataset-feaf1b793c6d408aaec1cac7cc3ceadc.pb.gz	130	0	0
	1241	0	0