Add initial build info about prosit HOT 10 CLOSED

@MarcIsak @gessulat

Nvidia-docker is not supported under WINDOWS. Reason, it would require a pass-through of all the GPU functionality. See https://github.com/NVIDIA/nvidia-docker/wiki/Frequently-Asked-Questions#is-microsoft-windows-supported

Is Microsoft Windows supported?
No, we do not support Microsoft Windows (regardless of the version), however you can use the native Microsoft Windows Docker client to deploy your containers remotely (refer to the dockerd documentation).

Do you support Microsoft native container technologies (e.g. Windows server, Hyper-v)?
No, we do not support native Microsoft container technologies.

So I see multiple solutions.

1. Use the native "jump" code example to run prosit predictions and training under Windows. Basically remove the server functionality and just run a python training and prediction set under Windows using TensorFlow GPU, Keras, numpy and all other dependent packages.

2. Use dual-boot on a Windows machine, allowing a LINUX partition to be selected upon boot.

3. Use Amazon or Google cloud to run prosit

4. Use the web-service from the Kusterlab

5. Ask a Linux guy if they can install the prosit software

6. Get a "cheap" LINUX box with a GTX 1050 (they are around 600 dollars)

from prosit.

The output looks good I think. You can try make jump and see if you get an interactive shell within the built container.

from prosit.

Will add detailed build information!

from prosit.

Very useful. A more detailed guide in README should help people get started easily.

from prosit.

Please refer to: https://github.com/kusterlab/prosit#installation
The required software and its versions are stated as tested.

from prosit.

Hi, I have not managed to run prosit on my computer, but I guess I am quite close.

I am running bash on Windows 10 (I use the Windows subsystem for Linux - Ubuntu Distribution). I can access the prosit folder (which should contain all files plus installed files, see attached image). Unfortunately, it seems like prosit cannot find the nvidia-docker when I try to use the make server command(see error message in the attached bash screen dump

).

I am not sure if this is because I did not configure the dependencies properly). I have downloaded docker, nvidia-docker and make as specified, but I did it before installing prosit..

Any tips to a computer rookie?

from prosit.

I would assume that the error is caused by one of the following:

1. nvidia-docker is not installed
2. nvidia-docker is not in your path

To ensure that nvidia-docker is setup in a way so that Prosit's makefile can make use of it you can run

 nvidia-docker run --runtime=nvidia --rm nvidia/cuda:9.0-base nvidia-smi

Which should show you the current utilization of you GPU(s).

https://github.com/NVIDIA/nvidia-docker recommends

docker run --runtime=nvidia --rm nvidia/cuda:9.0-base nvidia-smi

which is basically the same.

from prosit.

The order in which you downloaded Prosit and it's dependencies does not matter. Except for nvidia-docker that needs to be installed after docker.

from prosit.

@tobigithub @gessulat

Hi,

yes, I realized that it would not work to install nvidia-docker on Windows. So instead I downloaded Ubuntu and ran it through Windows 10's WSL feature. Problem was that I couldn't install Docker properly as I got this message in the end when I wanted to see that everything worked out:
System has not been booted with systemd as init system (PID 1). Can't operate.

To address the above points.

1. Sounds interesting, but I have no clue how to set up this workaround and would need some help.
2. This is a viable option, that perhaps could help me get past the problem with installing Docker on the WSL Ubuntu.
3. No idea how to do this. Would need some help.
4. Maybe the last resort. Waiting times for result output is the problem here, and the limited number of RAW-file that can be uploaded.
5. Very likely
6. I think point 2 is a better option.

Thanks for your input! I won´t keep spamming about this so the thread is definitely closed here :).

from prosit.

@MarcIsak
actually the docker file is available, so under Windows
you can just run the pip installer or use a conda environment.

Then you modify the code without the docker "frame".
So usage under Windows should not be a problem.

FROM tensorflow/tensorflow:1.10.1-gpu-py3
RUN pip install keras==2.2.1 h5py tables flask pyteomics lxml

ENV KERAS_BACKEND=tensorflow
ENV TF_CPP_MIN_LOG_LEVEL=3
ENV LC_ALL=C.UTF-8
ENV LANG=C.UTF-8

ADD prosit/ /root/prosit
RUN cd /root/
WORKDIR /root/

from prosit.