📑 Feature Request: Improve Ersilia Documentation

[GemmaTuron]

Is your feature request related to a problem? Please describe.

Ersilia's Documentation could cover more comprehensively the installation issues users find, for example when installing the WSL if they are Windows users

Describe the solution you'd like.

The different contributors that have struggled by one reason or another to install ersilia should add a comment in this issue with the following structure:

Operating System: #all you can gather about your system if it is a windows, which windows and which wsl are you using Conda Version: Python Version: Issue:

detailed explanation of the problem you encountered and how did you realise it was this problem (was it thanks to an error log, was it a system notification...)

exact error notification that you got

How did you solve it: #how did you solve the issue explained What would you like to see in the Documentation:

which section of the documentation is more adequate to host this kind of issue

exact sentences that you would write in the docs to help the next user

What would you like Ersilia to throw back when this issue is encountered: #the error message that should be printed in the CLI

Please attach any error log files that you got.

Describe alternatives you've considered

No response

Additional context.

No response

[GemmaTuron]

This is an example, based on #277, #323 and #340

Operating System: Linux Ubuntu 20.4 LTS Conda Version: conda 4.11.0 (Anaconda 3) Python Version: Python 3.9.7 Issue: Git - LFS was not installed and therefore the model could not be fetched from the Git Repository At fetching time I get an "EmptyErrorOutput". In the error log I can see the following message git: 'lfs' is not a git command. See 'git --help'. (see attached log)

How did you solve it: I installed git-lfs using the command git-lfs install and re-started the system What would you like to see in the Documentation: In this case, the Git LFS requirement is already stated in the pre-requisites section. I would add a line that specifies the following "Upon installation of git-lfs through conda, you may need to restart the system so that the git-lfs package is recognised"

What would you like Ersilia to throw back when this issue is encountered: Warning: GIT-LFS is not installed

(I am not attaching any error files as this is only an example)

[Malikbadmus]

Operating system: Linux Ubuntu 22.4

Conda Version: Conda 22.9.0

Python version: Python 3.10.6

Issue: The Input file was not in my working directory and I did not specify a path to the file in my predict command I got a TypeError when I ran the predict command. I have attached an error log below. TypeError.log

How did you solve it: I move the input life into my current working directory, then ran the command again and it was a success!!.

What would you like to see in the Documentation: I noticed that there were a considerable amount of TypeError bug issues that were initiated by contributors. And this error was always as a result of the predict command not reading the input file properly and this can be as a result of.

1. Using an incorrect input file name in your predict command.
2. Not appending .csv in your file name.
3. Not specifying the correct path to the location of your input file.

Based on this, I would suggest adding a line in the contribution guidelines that specifies the following: "When running the predict command, make sure you specify a path to the prepared list of molecules you downloaded or you could move it to your Ersilia directory.

What would you like Ersilia to throw back when this issue is encountered: Warning: FileNotFoundError

[paulinebanye]

Operating System:

• Windows 10
• Ubuntu 20.04

Conda Version:

• 22.9.0

Python Version:

• version 3.9.12

Issue: I had an issue running conda activate ersilia at Pre-requisite 5: The Isaura data lake in the Ersilia book.

exact error notification that you got

I received an ersilia not found error in the terminal. I figured out that I had to go through the step of installing ersilia first before isaura.

What would you like to see in the Documentation: I would like the Pre-requisite 5: The Isaura data lake to be placed after 2.2 Install Ersilia.

which section of the documentation is more adequate to host this kind of issue

I would like this to be placed in the installation section

[ZakiaYahya]

Operating system: Windows 10 Pro (WSL installed with Ubuntu 18.04)

Conda Version: Conda 22.9.0

Python version: Python 3.7.13

Issue: While trying to install ersilia model hub, i was getting error continuously i.e.
Could not find a version that satisfies the requirement h5py==3.7.0 (from isaura) (from versions: none) ERROR: No matching distribution found for h5py==3.7.0. Warning: There was an error checking the latest version of pip I have attached screenshot of an error below.

How did you solve it: Do this inside the ersilia conda environment, sudo apt-get update sudo apt-get upgrade sudo apt install python3-pip pip3 --version
Deactivate the ersilia conda environment & Rerun the ubuntu terminal.

What would you like to see in the Documentation: I noticed that few of other contributors also face this error also while installation with different packages too e.g.
ERROR: Could not find a version that satisfies the requirement bentoml==0.11.0 (from ersilia) (from versions: none) ERROR: No matching distribution found for bentoml==0.11.0. (instead of h5py) , Warning: There was an error checking the latest version of pip This error occurs when pip is not installed properly inside the conda environment due to some reasons. so to avoid that error one should install the pip again inside the conda environment. Based on this, I would suggest adding a line that specifies the following: While installation of ersilia model hub, when someone get this error especially warning of "There was an error checking the latest version of pip"one should check the pip version installed in the conda environment or reinstalled the pip package again.

What would you like Ersilia to throw back when this issue is encountered: I would like this to be placed in the installation section

[ZakiaYahya]

Operating system: Windows 10 Pro (WSL installed with Ubuntu 18.04)

Conda Version: Conda 22.9.0

Python version: Python 3.7.13

Issue: When i tried fetching ersilia model then somewhere while installing dependancies of that model my windows crashes. Although when i run fetch api everything was working fine and dependencies were installing too but suddenly my windows crashes and through me an error, here's the screenshot of error:

How did you solve it: This happens because of Antivirus that blocks the WSL to install some dependencies in ubuntu folder created in your windows. So, i resolved this issue by uninstalling the Antivirus for time being and everything works fine. But you can also put the WSL files in Antivirus exception list to avoid windows to crash.

What would you like to see in the Documentation: Based on this, I would suggest adding a line that specifies the following: If you are a window's user using WSL and encounter windows crashing while running fetch api then one should try to either put the WSL files in exception list of their Antivirus or completey delete it.

What would you like Ersilia to throw back when this issue is encountered: I would like this to be placed in the "Windows User section" who are using WSL for Ersilia model hub

[paulinebanye]

Operating System:

• Windows 10
• Brave browser

Conda Version:

• 22.9.0

Python Version:

• version 3.9.12

Issue: I encountered a key error on colab, whilst converting the model prediction output - eos6o0z.csv to a smiles list.

How did you solve it: It was referencing CAN_SMILES which wasn't present in the input file. So I changed the smiles = df["CAN_SMILES"].tolist() to smiles = df["can_smiles"].tolist()

What would you like to see in the Documentation: I would like the template to be adjusted to reflect the correct column name.

What would you like Ersilia to throw back when this issue is encountered: Invalid Column name

link to issue

[paulinebanye]

Operating System:

• Windows 10

Conda Version:

• 22.9.0

Python Version:

• version 3.9.12

Issue: I encountered a type error whilst attempting to calculate the molecular weight of the test model eos3b5e.

TypeError: No registered converter was able to produce a C++ rvalue of type std::basic_string<wchar_t, std::char_traits, std::allocator > from this Python object of type NoneType

How did you solve it: I adjusted the input from the model name to the single molecule of 6 carbons ersilia -v api calculate -i "CCCCCC"

What would you like to see in the Documentation: I would like the instruction for the test model to reflect the carbon input instead of the model name.

link to issue

[Motunrayo321]

Operating System: Windows 10 Pro (WSL installed with Ubuntu 22.04)

Conda Version: conda 4.11.0 (miniconda 3)

Python Version: Python 3.9.9

Issue: ersilia: command not found

I had closed the terminal previously and picked off from where I left after reopening it. No ersilia command worked after I went back to the directory path.

How did you solve it: I activated the ersilia environment using the command: conda activate ersilia

What would you like to see in the Documentation: It was stated in the installation guide to set up a conda enivronment and I neglected to do that. However, as a reminder I would appreciate if it could state that this was a requirement every time, for the sake of newcomers.

What would you like Ersilia to throw back when this issue is encountered: Warning: Ersilia is not activated! Try: conda activate ersilia

[Motunrayo321]

Operating System: Windows 10 Pro (WSL installed with Ubuntu 22.04)

Conda Version: conda 4.11.0 (miniconda 3)

Python Version: Python 3.9.9

Issue: While trying to install WSL on my Windows 10 pro machine with the command "wsl --install", I received an error message: WslRegisterDistribution failed with error: 0x80370102 Error: 0x80370102 The virtual machine could not be started because a required feature is not installed.

How did you solve it: First, I checked the requirements for installing wsl on my computer and saw this: Hyper-V Visualisation enabled in firmware No

This indicated that Hardware visualisation was disabled in my computer. Then I tried a series of methods to turn it on and succeeded by restarting my computer and booting it in BIOS mode. After that, I enabled "Virtualisation technology" in the start up options and rebooted my computer.

What would you like to see in the Documentation: I would like a hint section that states: "If you encounter Error: 0x80370102 while trying to install wsl, it is likely that your computer does not meet the requirements for the installation. Try enabling Hardware Virtualisation on your system."

[paulinebanye]

Operating System:

• Windows 10
• Ubuntu 20.04

Conda Version:

• 22.9.0

Python Version:

• version 3.9.12

Issue: I experienced an EmptyOutputError while fetching models with the fetch command ersilia -v fetch eos157v

How did I solve this. I reported the bug and the mentors rectified the problem. I simply had to pull the latest changes from the repo and execute the fetch command again.

What would you like to see in the Documentation I think the emptyoutputerror message would suffice, however I would like an additional message that indicates that this error is not a user related error

link to issue

[paulinebanye]

Operating System:

• Windows 10
• Brave browser 1.44.105

Conda Version:

• 22.9.0

Python Version:

• version 3.9.12

Issue: I received this error while fetching model eos85a3 on collab Exception: eos85a3 environment does not exist.

How did you solve it: I reconnected the runtime environment and that resolved the issue.

What would you like Ersilia to throw back when this issue is encountered: Environment does not exist due to a runtime error. Reconnect google colab.

link to issue

[paulinebanye]

Operating System:

• Windows 10
• Brave browser 1.44.105

Conda Version:

• 22.9.0

Python Version:

• version 3.9.12

Issue/Improvement: This isn't really a bug I encountered. It's more like a feature which would enhance usability of the colab environment.

A a first time user to google collab, I struggled with identifying the correct path for the input file. I eventually figured it out but I noticed that several contributors on the workspace encountered the same issue.

How would a user solve it:

• [x] Download the input file to your local system.
• [x] Connect to Google drive
• [x] Click on the "files" icon on the navigation bar, to the right of the collab window (it's the last icon)
• [x] Identify the folder you intend to store the input file.
• [x] Click on the three dots to the left of the folder name and select "Upload".
• [x] Navigate to and select the downloaded input file.
• [x] Once uploaded, right click and select "copy path".
• [x] Paste this path in the string to specify the path to the input file on colab.

What would you like to see in the Documentation: A section on testing the models using colab.

What would you like Ersilia to throw back when this issue is encountered: Follow the steps to upload your input file

[Femme-js]

Operating system: Ubuntu 18.04.3 LTS

Conda Version: conda 4.12.0

Python version: Python 3.9.12

Issue: I encountered a TypeError while predicting the eml_canonical.csv.

How did you solve it: I mentioned the exact path of the input file in the predict command and executed the command. It ran successfully.

What would you like to see in the Documentation: As new contributors or people who are just starting out with open-source contributions and CLI, in general, might face this issue where predict command fails to read the input file.

I would suggest adding a precautionary statement in the documentation, that specifies the TypeError that can occur due to the incorrect file path/format.

This can be similar to the already present precautionary statement in the ersilia's documentation regarding the available file formats.

What would you like Ersilia to throw back when this issue is encountered:

Warning: FileDoesNotExist

[Cee-tech21]

https://github.com/ersilia-os/ersilia/issues/342

Operating System: Linux 19 Conda Version: conda 22.9.0 Python Version: Python 3.9.13

Issue: Repeated output of "status 500" error code. During prediction using a model, the process seems to get stuck in a loop and repeatedly outputs "status 500" error code: see #342.

How did you solve it:

• Run prediction for one molecule
• If prediction works for one molecule, run prediction for three molecules
• If prediction for three molecule works, then run prediction for the input file containing all your molecules

What would you like Ersilia to throw back when this issue is encountered:

• A message such as:

Try running prediction for one molecule only and see if prediction works. If prediction works for one molecule but failed with input file containing all (many) molecules, try reducing molecules contained in your input file.

[zheeburg]

Operating System: 
Linux Ubuntu 20.4.1 LTS


Conda Version:
 conda 22.9.0 


Python Version:
 Python 3.9.12


Issue:
 My windows crashes and restarted, stopping all active processes whenever i fetch the ersilia model.

How did you solve it:
 It turned out that systemd-oomd (a userspace out-of-memory (OOM) killer) was killing my applications whenever I was running low on swap space.

You can run the following command to see if it is the case for you as well: journalctl -u systemd-oomd Two options:

1. Disable systemd-oomd:
systemctl disable --now systemd-oomd

2. Increase swap size (recommended).

[zheeburg]

Operating System:
 Linux Ubuntu 20.4.1 LTS


Conda Version: 
conda 22.9.0 


Python Version:
 Python 3.9.12

ISSUE ENCOUNTERED: Whenever I tried fetching my model, I got this message /usr/lib/python2.7/site-packages/requests/init.py:91: RequestsDependencyWarning: urllib3 (1.25.2) and chardet (3.0.4) doesn't match a supported version!
RequestsDependencyWarning)

HOW TO FIX: You can fix this problem with;
 pip install --upgrade requests==2.20.1

If you see version incompatible message like following, you should try other versions. All versions are: here

If that doesnt work, you can try removing docker compose;
 sudo apt-get remove docker-compose and installing: sudo curl -L The versions are here "https://github.com/docker/compose/releases/download/1.24.0/docker-compose-$(uname-s)-$(uname-m)"-o/usr/local/bin/docker-compose

If that doesnt work, try downloading docker-compose directly;
 sudo curl -L "https://github.com/docker/compose/releases/download/1.24.0/docker-compose-$(uname-s)-$(uname-m)"-o/usr/bin/docker-compose

Then adding execution capability to the file by; sudo chmod +x /usr/bin/docker-compose I am sure this fixes your issues.

[zheeburg]

Operating System:
 Ubuntu 20.04.1 LTS

Conda Version: 
conda 22.9.0 


Python Version:
 Python 3.9.12

The Issue: Issue #349 I got an Error .CSV not found while predicting my models using the downloaded drive file “eml_canonical.csv”

How did you solve it: I found out that I didn’t add the .csv to the input file in my predict command, So I fixed that by correcting the name of the input file from “eml_canonical” to “eml_canonical.csv” in the predict command and run it again. Then it worked and I continued fetching using the right commands.

ersilia -v fetch eos7w6n
ersilia -v serve eos7w6n
ersilia -v api predict -i eml_canonical -o eos7w6n
.csv

What would you like to see in the Documentation: For individuals who could confront this issue where the command does not read the file I will recommend adding a prudent assertion in the documentation, that determines the possible problem that can happen because of the erroneous file type.

What would you like Ersilia to throw back when this issue is encountered: Error: Try running prediction using the correct file name/format.

[Malikbadmus]

Operating Environment Google Colab

Conda Version: Conda 4.2.0

Python version: Python 3.7.13

Issue: It was observed that the fetching of a new model immediately after successfully running another model on google colab raises the error Exception: (model name) environment does not exist.

This would imply that the current working session has to be terminated and Ersilia Re-installed on Google Collab for every Model Instance.

How did you solve it:

Seeing that the model ran successfully with the same block of codes on a new notebook that was created , I figured the issue might stem from the open sessions.

Disconnecting and deleting runtime will fix this issue.

What would you like to see in the Documentation:

While this might be a temporary Bugfix, I might suggest adding this script to terminate the session at the end of the lines of code on google colab.

from google.colab import runtime runtime.unassign()

[Femme-js]

Operating Environment Google Colab

Python version: 3.7.15

Issue: It was observed that while making predictions using predict API on CLI using an input file, we do not get a message indicating the total time taken to make predictions as we use to get while fetching the model. Even in the colab template, we do not get any printed messages regarding the total time taken by the model to predict.

How did you solve it: In CLI, we can note the start time of the log and end time, to calculate the time taken in prediction. In google colab, we can check the time taken by the cell to execute the predict command.

Another approach that I use for better readability is using the time module in python:

from ersilia import ErsiliaModel import time model = ErsiliaModel("eos2thm") model.serve() begin = time.time() output = model.predict(input=smiles, output="pandas") end = time.time() print(end - begin) model.close()

What would you like to see in the Documentation:

We can try to add this functionality in the code itself, to produce the total prediction time taken by the model. In the documentation, we can edit the collab template for better readability.

[EstherIdabor]

Operating Environment Windows 10 Ubuntu 18.04.5

Conda Version 22.9.0

python version 3.7.13

Issue After installing ersilia model hub, ersilia --help threw an error which indicates that ersilia was not properly installed, the error is a ContextualVersionConflict which arose because the packages needed to run ersilia are not the versions required, in other words are not available.

How did you solve it I started the installation from the beginning and ensure no step was missed and then in addition ran the following comands; sudo apt-get update, sudo apt-get upgrade, sudo apt-install python3-pip pip3 --version; to check its the latest version and then I reran the terminal

What would you like to see in the Documentation I would love to see that pip should be up to date to ensure there is no version conflict while running ersilia

What would you like Ersilia to throw back when this issue is encountered Error: ContextualVersionConflict(Ensure pip is up to date)

[alaminumar]

Operating System: Linux Ubuntu 20.4 LTS Conda Version: conda 4.11.0 (Anaconda 3) Python Version: Python 3.9.12

Issue: Ubuntu was not able to work after WSL installation in Windows 10. This is because Hyper V a virtualization package does not come preinstalled in windows 10 Home and Virtualization was disabled in BIOS.

How did you solve it: Enabled Virtualization in BIOS googled and installed Hyper V re-started the system

What would you like to see in the Documentation: A section showing how the issue can be resolved in the installation documentation

What would you like Ersilia to throw back when this issue is encountered: Warning: Virtualization is not enabled check

Additional information for further guidance. https://www.itechtics.com/enable-hyper-v-windows-10-home/

[carcablop]

Operating System: Ubuntu 20.4 LTS Conda Version: conda 22.9.0(Anaconda 3) Python Version: Python 3.7.13

Issue: When I fetch a model it hosts an error in the log: git: 'lfs' is not a git command. See 'git --help'. (see attached log) eos526j_fetch_5.log Despite following the installation steps and reinstalling everything, the error still persists, as described in the bug #376

How did you solve it: sudo apt install git-lfs git-lfs install and re-started the system

What would you like to see in the Documentation:

I would like another option to install git lfs in the part of prerequisite 4: Git LFS. I would add a new line that specifies :" This can also be installed like this: sudo apt install git-lfs. Restart the system, verify that git-lfs is installed with following command: git-lfs --version.

What would you like Ersilia to throw back when this issue is encountered: Error: GIT-LFS is not installed

[carcablop]

Operating System: Ubuntu 20.4 LTS Conda Version: conda 22.9.0(Anaconda 3) Python Version: Python 3.7.13

Issue: When trying to fetch a model I got this error: TypeError: 'NoneType' object is not subscriptable. Reviewing the log I found that an ErrorException had been generated

In the previous line you can see that the download of torch was not completed since the system storage was too full to continue downloading torch.

How did you solve it: As I use an Ubuntu running on a virtual box machine, I increased the storage size of my machine, this was done in the virtual box configuration. After booting Ubuntu, since the sda partition was not the right size, I resized the partition to the size of the disk. I restarted the system and I already had the space I needed free.

What would you like to see in the Documentation: Specify the system requirements, such as minimum disk storage and minimum ram memory for fetch serve and predict the models.

What would you like Ersilia to throw back when this issue is encountered: In this case I would like to see the following in the error description: Error: torch did not complete its download, please check if you have enough space on your system." Other option: Error: Download not complete .Please check your system disk space. my.log

[Zainab-ik]

Operating System:

Windows 11 - Linux Ubuntu 20.4 LTS

Conda Version:

conda 22.9.0 (Miniconda)

Python Version:

Python 3.7.13

Issue/Improvement:

This is a feature in colab to enhance the user experience. While using the model, Google colab has a timestamp feature that records the duration of time for each cell code run. This could be used to record fetch and predict time as calculating the time code might come up with errors.

How did you solve it:

Simply enable the status bar feature at the lower part of the colab page to see the duration of time for the process to run. This changes for each code cell. Also, the log output in colab actually state the duration which should be taken into account.

Screenshot

What would you like to see in the Documentation:

This a reminder to check the status bar for Fetch time and predict time.

What would you like Ersilia to throw back when this issue is encountered: This is not an issue to be precise. However, after the model successfully fetches or predicts, I'd like ersilia to throwback thecheck status bar for the duration of the process after the model fetched succesfully.

[Zainab-ik]

Operating System:

Windows 11 - Linux Ubuntu 20.4 LTS

Conda Version:

conda 22.9.0 (Miniconda)

Python Version:

Python 3.7.13

Issue/Improvement:

While installing the ersilia python package by running the clone command git clone https://github.com/ersilia-os/ersilia.git to clone the ersilia repository to my local machine. The following error showed:

error: RPC failed; curl 92 HTTP/2 stream 5 was not closed cleanly before end of the underlying stream error: 2427 bytes of body are still expected fetch-pack: unexpected disconnect while reading sideband packet fatal: early EOF fatal: fetch-pack: invalid index-pack output

see issue #296

How did you solve it:

I simply noticed I was running the command in a different Operating System as opposed to the OS compatible with Ersilia which is WSL for windows. I changed to ubuntu to rerun the command once again and it worked.

Screenshot

Error: Solution

What would you like to see in the Documentation:

A reminder to always run the ersilia code in a Linux environment for Windows users, a step-by-step guide to enable WSL on windows or an additional link for the procedure. Also, an improved feature for ersilia to be able to run on windows.

What would you like Ersilia to throw back when this issue is encountered:

OS:Error, Verify OS to tell users they might be running their commands in the wrong environment.

[Zainab-ik]

Operating System:

Windows 11 - Linux Ubuntu 20.4 LTS

Conda Version:

conda 22.9.0 (Miniconda)

Python Version:

Python 3.7.13

Issue/Improvement:

While fetching the model in Colab, it returned an error Exception: Download from ersilia-os/eos3d3r did not work when it's supposed to fetch succesfully.

How did you solve it:

I figured that there was an internet break along the line and increased the broadband for my wifi to avoid an internet break. I also deleted and disconnect runtime before running again.

What would you like to see in the Documentation:

Ersilia works error freely while connected to the Internet

What would you like Ersilia to throw back when this issue is encountered:

Internet connection error: Make sure you are connected to internet

[Zainab-ik]

Operating System:

Windows 11 - Linux Ubuntu 20.4 LTS

Conda Version:

conda 22.9.0 (Miniconda)

Python Version:

Python 3.7.13

Issue/Improvement:

While installing ersilia using the pip install -e, it returned a no matching distribution .

How did you solve it:

I figured that there was an internet break along the line and increased the broadband for my wifi to avoid an internet break. I re-activated the ersilia environment in my terminal.

What would you like to see in the Documentation:

Ersilia works error freely while connected to the Internet. I'd like to state that this error persists and sometimes it's not due to an internet problem. Ersilia's debug feature should be able to classify this error as either an internet error or not. An internet breakout could occur while downloading some packages and it could be that the version the model is downloading is not compatible. However, these scenarios return the same error message and make it not so easy to debug.

What would you like Ersilia to throw back when this issue is encountered:

Due to an internet error: Internet connection error: Make sure you are connected to the internet Due to package incompatibility: contradicting package version. Suggest updating of packages.

[zheeburg]

Operating System:
 Linux Ubuntu 20.4 LTS


Conda Version:
 conda 22.9.0 


Python Version: 
Python 3.9.12


Issue:
 When I was installing some new packages (actually after trying to install build-essential), I got the error, Package installation fails with 404 errors.

Here is the error $ sudo apt install build-essential Reading package lists... Done Building dependency tree Reading state information... Done The following additional packages will be installed: cpp cpp-9 dpkg-dev fakeroot g++ g++-9 gcc gcc-9 gcc-9-base libalgorithm-diff-perl libalgorithm-diff-xs-perl libalgorithm-merge-perl libasan5 libatomic1 libc-dev-bin libc6-dev libcc1-0 libcrypt-dev libdpkg-perl libfakeroot libfile-fcntllock-perl libgcc-9-dev libisl22 libitm1 liblsan0 libmpc3 libquadmath0 libstdc++-9-dev libtsan0 libubsan1 linux-libc-dev make Suggested packages: cpp-doc gcc-9-locales debian-keyring g++-multilib g++-9-multilib gcc-9-doc gcc-multilib autoconf automake libtool flex bison gdb gcc-doc gcc-9-multilib glibc-doc bzr libstdc++-9-doc make-doc The following NEW packages will be installed: build-essential cpp cpp-9 dpkg-dev fakeroot g++ g++-9 gcc gcc-9 gcc-9-base libalgorithm-diff-perl libalgorithm-diff-xs-perl libalgorithm-merge-perl libasan5 libatomic1 libc-dev-bin libc6-dev libcc1-0 libcrypt-dev libdpkg-perl libfakeroot libfile-fcntllock-perl libgcc-9-dev libisl22 libitm1 liblsan0 libmpc3 libquadmath0 libstdc++-9-dev libtsan0 libubsan1 linux-libc-dev make 0 upgraded, 33 newly installed, 0 to remove and 33 not upgraded. Need to get 1096 kB/35.2 MB of archives. After this operation, 157 MB of additional disk space will be used. Do you want to continue? [Y/n] y Ign:1 http://archive.ubuntu.com/ubuntu focal-updates/main amd64 linux-libc-dev amd64 5.4.0-31.35 Err:1 http://security.ubuntu.com/ubuntu focal-updates/main amd64 linux-libc-dev amd64 5.4.0-31.35 404 Not Found [IP: 91.189.88.142 80] E: Failed to fetch http://security.ubuntu.com/ubuntu/pool/main/l/linux/linux-libc-dev_5.4.0-31.35_amd64.deb 404 Not Found [IP: 91.189.88.142 80] E: Unable to fetch some archives, maybe run apt-get update or try with --fix-missing?

The important bit: Failed to fetch http://security.ubuntu.com/ubuntu/pool/main/l/linux/linux-libc-dev_5.4.0-31.35_amd64.deb 404 Not Found [IP: 91.189.88.142 80]

Running with --fix-missing as suggested in the error message and a few blog posts gave the same error. sudo apt-get update didn't work either:

$ sudo apt-get update --fix-missing && apt-get install gcc Hit:1 https://dl.yarnpkg.com/debian stable InRelease Hit:2 http://archive.ubuntu.com/ubuntu focal InRelease Hit:3 https://deb.nodesource.com/node_12.x focal InRelease Get:4 http://security.ubuntu.com/ubuntu focal-security InRelease [107 kB] Get:5 http://archive.ubuntu.com/ubuntu focal-updates InRelease [111 kB] Get:6 http://archive.ubuntu.com/ubuntu focal-backports InRelease [98.3 kB] Reading package lists... Done E: Release file for http://security.ubuntu.com/ubuntu/dists/focal-security/InRelease is not valid yet (invalid for another 10d 6h 3min 42s). Updates for this repository will not be applied. E: Release file for http://archive.ubuntu.com/ubuntu/dists/focal-updates/InRelease is not valid yet (invalid for another 10d 6h 3min 50s). Updates for this repository will not be applied. E: Release file for http://archive.ubuntu.com/ubuntu/dists/focal-backports/InRelease is not valid yet (invalid for another 10d 6h 4min 14s). Updates for this repository will not be applied.

How did you solve it: The issue was that my clock was out of sync with the Windows host (I don't know why). So I used this code; $ sudo hwclock --hctosys

After this, $ sudo apt update ran happily, as did $ sudo apt install build-essential.

[GemmaTuron]

Thanks for all the contributions, you can keep adding them in this issue! Meanwhile we will start collecting the errors into a comprehensive documentation, separated by different issue types.

@zheeburg has offered to start with the documentation around installing the WSL. I suggest @zheeburg you prepare a GDocs and share it here so people can add comments and suggestions. Once the doc is finished we will incorporate it in the Ersilia Documentation

[zheeburg]

Thanks for all the contributions, you can keep adding them in this issue! Meanwhile we will start collecting the errors into a comprehensive documentation, separated by different issue types.

@zheeburg has offered to start with the documentation around installing the WSL. I suggest @zheeburg you prepare a GDocs and share it here so people can add comments and suggestions. Once the doc is finished we will incorporate it in the Ersilia Documentation

Okay Sure, Thanks I will do that now.

[zheeburg]

Hi Everyone I created a detailed guide on how to install WSL2 on different versions of windows 10 and I added an issue already, also how I fixed it. I’d love to have your comments on it so we can create a detailed document. Here is the link to the Google Docs.

https://docs.google.com/document/d/13qddcK8_Lfe-6qbNhAdTJfXIVE6PckRuuWuLrKrclEE/edit

[Femme-js]

@GemmaTuron

I also encountered an error while installing Ersilia with the help of instructions. But I quickly figured it out given that I already have experience when dealing with virtual environments. But later I realized a lot of folks went through the same issue.

Mentioning this, I would like to assist you with improving the installation guidelines when working with Ersilia on Linux.

[Chigoziee]

Operating System: Windows 10 Pro (WSL installed with Ubuntu 22.04)

Conda Version: conda 22.9.0 (Miniconda)

Python Version: Python 3.9.12

Issue: After the installation of ubuntu and Ersillia (and all other additional packages). I was stuck on a virtual environment so I could not load my test data into the model because I could not locate the directory which it was saved at.

How did you solve it: I went to my Ubuntu settings and changed my Ubuntu starting directory from the wsl virtual environment to my desktop this helped me to easily navigate my PC. in doing that I had to reinstall Ersilia again; start over the entire installation process which then allowed everything to work perfectly.

What would you like to see in the Documentation: i would like for it to be stated that after installing ubuntu, windows users should ensure that their starting directory is in their desktop (or at least a directory they can identify) before installing ersilia.

I would like a hint section that states: i was only able to locate this issue because my test data was not been found. i would like for a more descriptive error like 'FileNotFound: check if file is at the current working directory'

[paulinebanye]

Operating System: Windows 10 Brave browser 1.44.105

Conda Version: 22.9.0

Python Version: version 3.9.12

Issue: It was observed that the instructions for making model predictions on colab was not explanatory or user friendly.

How did you solve it: To resolve this, I started by incorporating feedback from other contributors @Malikbadmus @ZakiaYahya @Femme-js and edited the Ersilia colab template by:

• [x] Assigned the model, input csv file, input & output folders as variables
• [x] Separated the fetch, serve and predict/calculate functions into distinct code blocks to output the return and for readability.
• [x] Including the time functionality suggested by @Femme-js on the fetch function to the predict/calculate function
• [x] Updated the comments and documentation for each code block.

What would you like to see in the Documentation: In the guide, we can update the information for more explicit information i.e. the logs

Link to colab template

[UChisom]

Operating System: # Windows 10, WSL Ubuntu 22.04.1 LTS Conda Version: 22.9.0 Python Version: 3.7.13

Issue: Virtualization Technology was disabled on my windows PC

The virtualization and Hyper-V were disabled by default on my Windows PC. Ubuntu repeatedly failed to install (right from the Microsoft store) due to this as it couldn't be hosted virtually upon the Windows OS. I opened Task Manager and checked the Performance tab to find the disabled Virtualization feature.

How did you solve it:

See this link to Microsoft's step-by-step guide to turn on the features

• Navigate to Turn Windows Features on/ off and enable Hyper-V checkbox
• Enabled virtualization technology from the BIOS set up when booting system (see link for HP)

What would you like to see in the Documentation:

What part of the docs would you like to see it?

OS Installation Guide

Exact words

Ensure that virtualization and Hyper V are enabled on your PC if on Windows

What would you like Ersilia to throw back when this issue is encountered:

Ersilia's setup would be incomplete at this point, so nothing

[UChisom]

Operating System: #all you can gather about your system if it is a windows, which windows and which wsl are you using Conda Version: Python Version:

Issue: Cloning the Isaura repository - While running the commands as stated in the Ersillia installation guide, the command git clone https://github.com/ersilia-os/isaura.git threw an error back and the clone was unsuccessful

How did you solve it:

I ran the command in the Command Line Interface: sudo apt git clone https://github.com/ersilia-os/isaura.git and it worked

What would you like to see in the Documentation:

#which section of the documentation is more adequate to host this kind of issue Isaura installation guide

#exact sentences that you would write in the docs to help the next user

This may just have been unique to my setup for Ubuntu, but a note would be: try also sudo apt git clone https://github.com/ersilia-os/isaura.git

[UChisom]

Operating System: Windows 10 Environment: Google Colab Python Version: 3.7.13

Issue: ersilia serve not specified from format Ersilia's colab format is found in Github here

The ersillia model serve command is embedded in the code below:

model = ErsiliaModel("eos2gth")
model.serve()
output = model.predict(input=smiles, output="pandas")

This hides the output from the model.serve() command and user won't know the available API

How did you solve it:

I broke the code block into 3 parts, having the model.serve() in a code cell so I can see what API is available for the model

What would you like to see in the Documentation:

I would like the separation in the code block to be made in the final documentation so the work flow is more seamless

[RoseZhen2022]

Operating System: Ubuntu 22.04.1 LTS, google colab Conda Version:22.9.0 Python Version:Python 3.7.13 Issue: Model can be fetched successfully. However, when doing prediction, it just keep running till time out. Then I need to fetch it again and do the prediction and still keep running. I couldn't get any error message, it seems the prediction didn't finish.

How did you solve it: Change the input file to a single molecule, then it can predict successfully. It means the model works well, but it needs a lot of time to predict. output = model.predict(input='CCCO', output="pandas")

What would you like to see in the Documentation: If someone encounter similar situation like when the model do predicting or calculating, it keeps running without any error message, you can try just input one single molecule to test whether the model can work well or not.

What would you like Ersilia to throw back when this issue is encountered: when the model executed exceed some specific time, it should be printed some message in the CLI

Additional context. The model I tested can be successfully predicted in my local environment, it needs more than 6 hours. So it may be due to time, but I hope it can give some hints.

[Zainab-ik]

Hi Everyone I created a detailed guide on how to install WSL2 on different versions of windows 10 and I added an issue already, also how I fixed it. I’d love to have your comments on it so we can create a detailed document. Here is the link to the Google Docs.

https://docs.google.com/document/d/13qddcK8_Lfe-6qbNhAdTJfXIVE6PckRuuWuLrKrclEE/edit

@zheeburg Let's work together. I use Windows 11 and took note of different instances in case of future users using the same version.

[zheeburg]

Hi Everyone I created a detailed guide on how to install WSL2 on different versions of windows 10 and I added an issue already, also how I fixed it. I’d love to have your comments on it so we can create a detailed document. Here is the link to the Google Docs. https://docs.google.com/document/d/13qddcK8_Lfe-6qbNhAdTJfXIVE6PckRuuWuLrKrclEE/edit

@zheeburg Let's work together. I use Windows 11 and took note of different instances in case of future users using the same version.

Of course, Thanks so much. We can

[muskansawa]

Operating System: Ubuntu 20.4 LTS (WSL) Conda Version: conda 22.9.0 Python Version: Python 3.7.13

Issue:

Unable to fetch model on running ersilia -v fetch eos2mrz because of a traceback error, On checking if all the requirements are installed or not in my system or not, I found out that docker was not installed in my WSL.

How did you solve it:

The problem was that I only had docker for my windows system and not for ubuntu. so I first Enable WSL 2 feature on Windows for ubuntu and then enabled the WSL 2-based engine in my docker desktop following the steps mentioned here: https://docs.docker.com/desktop/windows/wsl/

What would you like to see in the Documentation:

I would like to see the above steps mentioned in the installation part in the section which is dedicated for windows system.

What would you like Ersilia to throw back when this issue is encountered: Warning: docker is not installed

[mahamtariq58]

System

Ubuntu 22.04.1 LTS Python 3.9 Conda 22.9.0

Issue:

Unable to fetch the model eos8d8a because the pip dependency resolver does not take into account all the packages that are installed.

How do I solve it:

• Check the closed issues, if anyone has also encountered the same issue
• I find one closed issue and I follow the instructions
• Install the Python 3.7 version

What would you like to see in the documentation

Based on the above mention issue, It would be better to explicitly mention that users may encounter some issues if they use the other python version.

[carcablop]

Operating System: Linux Ubuntu 20.4 LTS Conda Version: conda 22.9.0(Anaconda 3) Python Version: Python 3.9.7 Issue: The Ubuntu system crashes when we try to make predictions with the eos526j model from Google Colab. The execution takes about 6 hours and the system hangs. How did you solve it: After rebooting the system, try to predict with a single molecule that we passed as input in the "predict" function, as can be seen in issue: #392

What would you like to see in the Documentation: Seeing that many of the contributors presented similar errors, without having a possible execution result, I consider that a session of possible failures in the models be added to the documentation and, in case they cannot be executed, show alternatives, in this case, add the suggestion of trying to make predictions with a molecule when the system collapses, in this way an investigation of the problem can be made.

What would you like Ersilia to throw back when this issue is encountered:

Ersilia should be able to show the execution time in the console and if it exceeds a very large limit in hours of execution, it should stop the prediction so as not to cause system failures. And at least show an output of the prediction of a molecule with its execution time.

[Zainab-ik]

Thanks for all the contributions, you can keep adding them in this issue! Meanwhile we will start collecting the errors into a comprehensive documentation, separated by different issue types.

@zheeburg has offered to start with the documentation around installing the WSL. I suggest @zheeburg you prepare a GDocs and share it here so people can add comments and suggestions. Once the doc is finished we will incorporate it in the Ersilia Documentation

Hi @GemmaTuron, that's great. I'd love to assist with the incorporation to Ersilia documentation.

[angelnagaba]

Operating System: Linux Ubuntu 22.04 LTS Conda Version: conda 22.9.0(Anaconda 3) Python Version: Python 3.9.12 Issue: Model doesn't run when you use a wrong API method for example running it with calculate instead of predict How did you solve it: I solved this by running the correct method. However it took me time to figure out since the error message wasn't clear

What would you like to see in the Documentation: This should be stated clearly in the documentation that users should pay attention to the Available API methods after serving a model

What would you like Ersilia to throw back when this issue is encountered:

A message that clearly shows why the model is not running. for example: "Wrong API method, check your available methods and run the model again"

[Celinezzzzz]

Operating System: Windows 10 WSL Ubuntu 20.04 LTS Conda Version: Miniconda3 Python Version: Python 3.9

Issue: I was testing models(eos3xip, eos77w8, eos5smc) on Colaboratory. I first test eos3xip and it worked successfully. When I was fetching the 2nd model, I got an error message "Exception: eos77w8 environment does not exist". I then tried it with the 3rd model eos5smc and the 1st model eos3xip, but I still got the same error message.

How did you solve it: I solved the issue by disconnecting and deleting runtime.

What would you like to see in the Documentation: But I am not sure what is causing this issue. Are we supposed to disconnect and delete runtme everytime after testing a model? I would explain this error in the Ersilia on Google Colab section and provide solution such as disconnecting and deleting runtime.

455

[sayantani11]

Operating System: Windows 11 WSL Ubuntu 18.04 LTS Conda Version: conda 22.9.0 Python: Python 3.7.13

Issue:

340

Initially the problem that was faced was EmptyOutputError but in line 9 & 10 of the log it was seen that the git-lfs is broken although I followed installation steps from the guide.

How did you solve it: Git LFS initialization had to be done even in ersilia enc created by conda. Having git-lfs in your local system only will show git-lfs as brolen.

What would you like to see in the Documentation: This should be mentioned in the installation steps that we also have to initialize git-lfs in our ersilia environment as well.

What would you like Ersilia to throw back when this issue is encountered: It would be great if the log can highlight or mark whenever it gets an ERROR message in a different color.

[Femme-js]

Operating system: Ubuntu 18.04.3 LTS

Conda Version: conda 4.12.0

Python version: Python 3.9.12

Issue: While working on Stage-3 contributions, I saw some gaps in Ersilia's documentation related to the QSAR modeling introduction and briefing about its AutoML tools. For new users or contributors, these additions can help them get onboard on the platform smoothly.

How did you solve it: I currently wrote a mock-up blog, which can be improved later on while integrating with Ersilia's gitbook.

Mock-up Blog: ''' QSAR Modelling with Ersilia

Quantitative structure-activity relationship models are the tools to predict the physicochemical and biological properties of chemical compounds from their chemical structure. QSAR modeling is one of the major computational in medicinal chemistry whose main goal is to establish a quantitative relationship between descriptors and the target property capable of predicting the activities of novel compounds.

Ersilia Open Sorce Initiative provides two major tools for Automate machine learning and on-demand modeling. ZairaChem for advanced and more accurate modeling, and LazyQSAR for quick baseline modeling.

ZairaChem is capable of producing State of the art results, without any need for data preprocessing and hyperparameter tuning. Model validation test results can be automatically served for better assessment. Lazy-QSAR is an AutoML library to build QSAR models fastly, and quickly build binary classification and regression models.

'''

What would you like to see in the Documentation: New contributors or people who are just starting out with open-source contributions and CLI, in general, might face difficulty in getting to know about Ersilia's tool without prior knowledge about QSAR modeling.

I would suggest adding a separate section or blog on the same in Ersilia's gitbook.

[GemmaTuron]

Thanks for the contributions from all Outreachy applicants! With @miquelduranfrigola and @Amna-28 we have been working on improving the exceptions module to better explain all those errors. Hopefully it will be much easier for future users! I'll close this issue but we will work with the Outreachy interns to transfer more information from here to the GitBook