Question about installation

[Yansr3]

I'm trying to create a mira environment following the steps in Getting Started. I'm using a M1 macos system. I tried with both python 3.11 and python 3.10, but both didn't work. May I ask for any help or suggestions on setting up the environment?

conda create --name mira-env -c conda-forge -c bioconda scanpy jupyter leidenalg conda activate mira-env conda install -c bioconda mira-multiome

But I'm getting this Unsatisfiable Error. (mira-env) ➜ ~ conda install -c bioconda mira-multiome Collecting package metadata (current_repodata.json): done Solving environment: unsuccessful initial attempt using frozen solve. Retrying with flexible solve. Solving environment: unsuccessful attempt using repodata from current_repodata.json, retrying with next repodata source. Collecting package metadata (repodata.json): done Solving environment: unsuccessful initial attempt using frozen solve. Retrying with flexible solve. Solving environment: | Found conflicts! Looking for incompatible packages. This can take several minutes. Press CTRL-C to abort. failed

UnsatisfiableError: The following specifications were found to be incompatible with each other:

Output in format: Requested package -> Available versions

[Yansr3]

I also tried to install with pip. But I got another conflict error. ERROR: Cannot install mira-multiome==0.0.0, mira-multiome==0.0.1, mira-multiome==0.0.2, mira-multiome==0.0.3, mira-multiome==0.0.4, mira-multiome==0.0.5, mira-multiome==0.0.6, mira-multiome==0.0.7, mira-multiome==0.0.8, mira-multiome==0.0.9, mira-multiome==0.1.0, mira-multiome==0.1.3, mira-multiome==1.0.0, mira-multiome==1.0.1, mira-multiome==1.0.2, mira-multiome==1.0.3, mira-multiome==1.0.4 and mira-multiome==2.1.0 because these package versions have conflicting dependencies.

The conflict is caused by: mira-multiome 2.1.0 depends on torch<2 and >=1.8.0 mira-multiome 1.0.4 depends on torch<2 and >=1.8.0 mira-multiome 1.0.3 depends on torch<2 and >=1.8.0 mira-multiome 1.0.2 depends on torch<2 and >=1.8.0 mira-multiome 1.0.1 depends on torch<2 and >=1.8.0 mira-multiome 1.0.0 depends on torch<2 and >=1.8.0 mira-multiome 0.1.3 depends on torch<2 and >=1.8.0 mira-multiome 0.1.0 depends on torch<2 and >=1.8.0 mira-multiome 0.0.9 depends on torch<2 and >=1.8.0 mira-multiome 0.0.8 depends on torch<2 and >=1.8.0 mira-multiome 0.0.7 depends on torch<2 and >=1.8.0 mira-multiome 0.0.6 depends on torch<2 and >=1.8.0 mira-multiome 0.0.5 depends on torch<2 and >=1.8.0 mira-multiome 0.0.4 depends on torch<2 and >=1.8.0 mira-multiome 0.0.3 depends on torch<2 and >=1.8.0 mira-multiome 0.0.2 depends on torch<2 and >=1.8.0 mira-multiome 0.0.1 depends on torch<2 and >=1.8.0 mira-multiome 0.0.0 depends on torch<2 and >=1.8.0

[AllenWLynch]

Hi, thanks for reaching out.

Would it be possible to try python 3.9 as well? I have found this issue and gotten access to an M1 mac so I can troubleshoot these dependencies. With the release of the M1 chips and updates to python I may need to bump some of the dependencies to a higher version.

AL

---

From: Yansr3 @.> Sent: Wednesday, August 23, 2023 2:30 PM To: cistrome/MIRA @.> Cc: Subscribed @.***> Subject: Re: [cistrome/MIRA] Question about installation (Issue #28)

I also tried to install with pip. But I got another conflict error. ERROR: Cannot install mira-multiome==0.0.0, mira-multiome==0.0.1, mira-multiome==0.0.2, mira-multiome==0.0.3, mira-multiome==0.0.4, mira-multiome==0.0.5, mira-multiome==0.0.6, mira-multiome==0.0.7, mira-multiome==0.0.8, mira-multiome==0.0.9, mira-multiome==0.1.0, mira-multiome==0.1.3, mira-multiome==1.0.0, mira-multiome==1.0.1, mira-multiome==1.0.2, mira-multiome==1.0.3, mira-multiome==1.0.4 and mira-multiome==2.1.0 because these package versions have conflicting dependencies.

The conflict is caused by: mira-multiome 2.1.0 depends on torch<2 and >=1.8.0 mira-multiome 1.0.4 depends on torch<2 and >=1.8.0 mira-multiome 1.0.3 depends on torch<2 and >=1.8.0 mira-multiome 1.0.2 depends on torch<2 and >=1.8.0 mira-multiome 1.0.1 depends on torch<2 and >=1.8.0 mira-multiome 1.0.0 depends on torch<2 and >=1.8.0 mira-multiome 0.1.3 depends on torch<2 and >=1.8.0 mira-multiome 0.1.0 depends on torch<2 and >=1.8.0 mira-multiome 0.0.9 depends on torch<2 and >=1.8.0 mira-multiome 0.0.8 depends on torch<2 and >=1.8.0 mira-multiome 0.0.7 depends on torch<2 and >=1.8.0 mira-multiome 0.0.6 depends on torch<2 and >=1.8.0 mira-multiome 0.0.5 depends on torch<2 and >=1.8.0 mira-multiome 0.0.4 depends on torch<2 and >=1.8.0 mira-multiome 0.0.3 depends on torch<2 and >=1.8.0 mira-multiome 0.0.2 depends on torch<2 and >=1.8.0 mira-multiome 0.0.1 depends on torch<2 and >=1.8.0 mira-multiome 0.0.0 depends on torch<2 and >=1.8.0

— Reply to this email directly, view it on GitHubhttps://github.com/cistrome/MIRA/issues/28#issuecomment-1690522763, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AE43JPFB3UBYVPPSAHRZOTLXWZK3TANCNFSM6AAAAAA336IX3M. You are receiving this because you are subscribed to this thread.Message ID: @.***>

[Yansr3]

Hi AL,

Thanks for responding. I tried with python 3.9 but neither conda nor pip install worked. I also tried python 3.5 to python 3.8, but they didn't work.

SY

[AllenWLynch]

Okay,

I have an M1 mac handy now. I will work on the dependencies and get back to you Monday-ish.

AL

---

From: Yansr3 @.> Sent: Thursday, August 24, 2023 2:58 PM To: cistrome/MIRA @.> Cc: AllenWLynch @.>; Comment @.> Subject: Re: [cistrome/MIRA] Question about installation (Issue #28)

Hi AL,

Thanks for responding. I tried with python 3.9 but neither conda nor pip install worked. I also tried python 3.5 to python 3.8, but they didn't work.

SY

— Reply to this email directly, view it on GitHubhttps://github.com/cistrome/MIRA/issues/28#issuecomment-1692325706, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AE43JPBWJQMWAXXKHEH6D3DXW6W53ANCNFSM6AAAAAA336IX3M. You are receiving this because you commented.Message ID: @.***>

[AllenWLynch]

Found the issue, now I am confirming that install works under more conditions. Today, I will submit pull requests to Bioconda to update the MIRA package dependencies, which will likely be ready tomorrow.

[Yansr3]

Thank you very much! I will try installing from bioconda tomorrow.

[AllenWLynch]

The issue was that the fast motif calling software I dependent on, MOODS, had some C compiler options which were not compatible on M1 Mac. The maintainer has not responded to an issue on this topic from three months ago, so I am fixing the compiler issues and distributing my own version. I need one more day to put it together, sorry for the inconvenience.

[Yansr3]

No worries, thank you very much for helping with this! I will wait and try again when it's ready.

[Yansr3]

Sorry this may be another question, but I want to quickly ask if it is possible to use GPU after I install mira into M1 Mac.

[AllenWLynch]

I believe not, I am unsure if Mac GPUs use CUDA. Would you be able to try installation with Pip instead of conda? The bioconda team is being slow approving my PRs, but I can have a pip update ready tomorrow.

AL

---

From: Yansr3 @.> Sent: Wednesday, August 30, 2023 10:18 AM To: cistrome/MIRA @.> Cc: AllenWLynch @.>; Comment @.> Subject: Re: [cistrome/MIRA] Question about installation (Issue #28)

Sorry this may be another question, but I want to quickly ask if it is possible to use GPU after I install mira into M1 Mac.

— Reply to this email directly, view it on GitHubhttps://github.com/cistrome/MIRA/issues/28#issuecomment-1699379713, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AE43JPCGEZIUZ4XICCXQWBTXX5KUHANCNFSM6AAAAAA336IX3M. You are receiving this because you commented.Message ID: @.***>

[AllenWLynch]

Hi, an update on running MIRA on python 3.10 and 3.11 systems. The bioconda team is being very sluggish approving my PRs still, but you can access a new alpha version for 3.10 and 3.11 on PyPi. To set up an env, run:

conda create --name mira-3.11 -c conda-forge -c bioconda scanpy jupyterlab ipywidgets seaborn
conda activate mira-3.11
pip install mira-multiome==2.1.1a3

This will undergo a bit more testing before full release. I have no idea how long it will take to get this hosted on Bioconda.

[Yansr3]

Hi, thank you very much for the updating! I succesfully installed with pip on python 3.9. I will also try on python 3.10 with the alpha veriosn.