502 Bad Gateway when accessing omnipath

[JohnGoertz]

Report

Via Python decoupler 2.1.4

Decoupler dc.op.show_resources() tries to access https://omnipathdb.org/queries/annotations, but that's been giving a "Bad Gateway" error for the past 24-48 hours.

This is called via dc.op.resource("MSigDB"). Is there an alternative (to access GO pathways) while Omnipath is down?

Versions

[asif7adil]

Facing same issue with
dc.op.progeny(organism="human")
Please fix or suggest an alternative!

Thanks

[OlivierBakker]

Just want to check if the following may be related related before I raise a new issue, I have been getting

ValueError: DataFrame constructor not properly called!

When trying to call: progeny = dc.op.progeny(organism="human")

This is using code and an enviroment that was working a few days ago.

collectri = dc.op.collectri(organism="human") does seem to run fine

p.s. also seems to come up here: #275

Stacktrace:

----> 1 progeny = dc.op.progeny(organism="human")

/lib/python3.10/site-packages/decoupler/op/_progeny.py in ?(organism, top, thr_padj, license, verbose)
     64     # Validate
     65     assert isinstance(top, int | float) and top > 0, "top must be numeric and > 0"
     66     assert isinstance(thr_padj, int | float) and 0.0 <= thr_padj <= 1.0, "thr_padj must be numeric and between 0 and 1"
     67     # Download
---> 68     p = resource(name="PROGENy", organism=organism, license=license, verbose=verbose)
     69     p = p.sort_values("p_value").groupby("pathway").head(top).sort_values(["pathway", "p_value"]).reset_index(drop=True)
     70     p = p.rename(columns={"pathway": "source", "genesymbol": "target", "p_value": "padj"})
     71     p = p[p["padj"] < thr_padj]

/lib/python3.10/site-packages/decoupler/op/_resource.py in ?(name, organism, license, verbose)
     75         df
     76     """
     77     # Validate
     78     assert isinstance(name, str), "name must be str"
---> 79     names = set(show_resources()["name"])
     80     assert name in names, f"name must be one of these: {names}"
     81     assert isinstance(organism, str), "organism must be str"
     82     assert isinstance(license, str) and license in ["academic", "commercial", "nonprofit"], (

/lib/python3.10/site-packages/decoupler/op/_resource.py in ?()
     31     ann = ann.set_index("argument").loc["databases"].str.split(";")["values"]
     32     url = "https://omnipathdb.org/resources"
     33     response = requests.get(url)
     34     lcs = response.json()
---> 35     df = pd.DataFrame(ann, columns=["name"])
     36     df["license"] = [lcs[a]["license"]["purpose"] if a in lcs else None for a in ann]
     37     return df

/lib/python3.10/site-packages/pandas/core/frame.py in ?(self, data, index, columns, dtype, copy)
    886                 )
    887         # For data is scalar
    888         else:
    889             if index is None or columns is None:
--> 890                 raise ValueError("DataFrame constructor not properly called!")
    891 
    892             index = ensure_index(index)
    893             columns = ensure_index(columns)

ValueError: DataFrame constructor not properly called!

[asif7adil]

Just want to check if the following may be related related before I raise a new issue, I have been getting

ValueError: DataFrame constructor not properly called!

When trying to call: progeny = dc.op.progeny(organism="human")

This is using code and an enviroment that was working a few days ago.

collectri = dc.op.collectri(organism="human") does seem to run fine

p.s. also seems to come up here: #275

Stacktrace:

----> 1 progeny = dc.op.progeny(organism="human")

/lib/python3.10/site-packages/decoupler/op/_progeny.py in ?(organism, top, thr_padj, license, verbose)
     64     # Validate
     65     assert isinstance(top, int | float) and top > 0, "top must be numeric and > 0"
     66     assert isinstance(thr_padj, int | float) and 0.0 <= thr_padj <= 1.0, "thr_padj must be numeric and between 0 and 1"
     67     # Download
---> 68     p = resource(name="PROGENy", organism=organism, license=license, verbose=verbose)
     69     p = p.sort_values("p_value").groupby("pathway").head(top).sort_values(["pathway", "p_value"]).reset_index(drop=True)
     70     p = p.rename(columns={"pathway": "source", "genesymbol": "target", "p_value": "padj"})
     71     p = p[p["padj"] < thr_padj]

/lib/python3.10/site-packages/decoupler/op/_resource.py in ?(name, organism, license, verbose)
     75         df
     76     """
     77     # Validate
     78     assert isinstance(name, str), "name must be str"
---> 79     names = set(show_resources()["name"])
     80     assert name in names, f"name must be one of these: {names}"
     81     assert isinstance(organism, str), "organism must be str"
     82     assert isinstance(license, str) and license in ["academic", "commercial", "nonprofit"], (

/lib/python3.10/site-packages/decoupler/op/_resource.py in ?()
     31     ann = ann.set_index("argument").loc["databases"].str.split(";")["values"]
     32     url = "https://omnipathdb.org/resources"
     33     response = requests.get(url)
     34     lcs = response.json()
---> 35     df = pd.DataFrame(ann, columns=["name"])
     36     df["license"] = [lcs[a]["license"]["purpose"] if a in lcs else None for a in ann]
     37     return df

/lib/python3.10/site-packages/pandas/core/frame.py in ?(self, data, index, columns, dtype, copy)
    886                 )
    887         # For data is scalar
    888         else:
    889             if index is None or columns is None:
--> 890                 raise ValueError("DataFrame constructor not properly called!")
    891 
    892             index = ensure_index(index)
    893             columns = ensure_index(columns)

ValueError: DataFrame constructor not properly called!

I think they're related, I was getting a bad gateway until yesterday, and I got the DataFrame constructor error today!

[PauBadiaM]

Hi @JohnGoertz @asif7adil and @OlivierBakker ,

Thanks for reporting this issue.

The error is caused by recent changes to the OmniPath server structure, which require corresponding updates in how decoupler queries it. I’ve started implementing a fix to accommodate these changes, but I’m currently blocked because the server appears to be down again.

@deeenes , it would have been helpful to coordinate these changes beforehand so I could have prepared and applied the necessary updates proactively. Could you please have a look at the server status?

Thanks!

[PauBadiaM]

As mentioned in #275, this issue has now been resolved. However, you may experience some download speed fluctuations over the coming weeks while the OmniPath team continues optimizing and managing the new server.

[asif7adil]

Working fine now!
Thanks!