Skip to content
Snippets Groups Projects
Commit 0e86b1fc authored by Paul McCarthy's avatar Paul McCarthy :mountain_bicyclist:
Browse files

Finished fsl-pipe section, added section on profiling and fsl_sub

parent a6d99654
No related branches found
No related tags found
1 merge request!38Add practical which introduces some third-party parallelisation libraries
Hide whitespace changes
Inline Side-by-side
applications/README.md
+ 1
− 0
View file @ 0e86b1fc
......@@ -12,3 +12,4 @@ Practicals are split into the following sub-categories (and sub-folders):
3. `matlab_vs_python` : a number of data analysis examples with head-to-head comparisons between matlab and python code.
4. `modelling` : multiple examples of analysis methods in python.
5. `pandas` : processing and analsing tabular data with the powerful Pandas package.
6. `parallel` : Third-party libraries and strategies for parallelising Python code.
applications/parallel/parallel.ipynb
+ 432
− 91
View file @ 0e86b1fc
%% Cell type:markdown id:2b88a4ad tags:
%% Cell type:markdown id:2b27adf9 tags:
# Parallel processing in Python
While Python has built-in support for threading and parallelising in the
[`threading`](https://docs.python.org/3/library/threading.html) and
[`multiprocessing`](https://docs.python.org/3/library/multiprocessing.html)
modules, there are a range of third-party libraries which you can use to improve the performance of your code.
modules (covered in `advanced_programming/threading.ipynb`), there are a range of third-party libraries which you can use to improve the performance of your code.
Contents:
* [Do you really need to parallelise your code?](#do-you-need-to-parallelise)
* [Profilng your code](#profiling-your-code)
* [JobLib](#joblib)
* [Dask](#dask)
* [Dask Numpy API](#dask-numpy-api)
* [Low-level Dask API](#low-level-dask-api)
* [Distributing computation with Dask](#distributing-computation-with-dask)
* [fsl-pipe](#fsl-pipe)
* [fsl-sub](#fsl-sub)
<a class="anchor" id="do-you-need-to-parallelise"></a>
# Do you really need to parallelise your code?
Before diving in and tearing your code apart, you should think very carefully about your problem, and the hardware you are using to run your code. For example, if you are processing MRI data for a number of subjects on the FMRIB cluster (where each node on the cluster has fairly modest hardware specs, meaning that within-process parallelisation will have limited benefits), the most efficient option may be to write your code in a single-threaded manner, and to parallelise across data sets, processing one data set on each cluster node.
Your best option may simply be to repeatedly call `fsl_sub`- see the [section below](#fsl-sub) for more details on this approach.
<a class="anchor" id="#profiling-your-code"></a>
## Profilng your code
Once you have decided that you need to parallelise some steps in your program, **STOP!** As the great Donald Knuth once said:
> Programmers waste enormous amounts of time thinking about, or worrying about, the speed of noncritical parts of their programs, and these attempts at efficiency actually have a strong negative impact when debugging and maintenance are considered. We should forget about small efficiencies, say about 97% of the time:
>
> **premature optimization is the root of all evil**.
>
> Yet we should not pass up our opportunities in that critical 3%.
What Knuth is essentially saying is that there is no point in optimising or rewriting a piece of code unless it is actually going to have a positive impact on performance. In other words, before you refactor your code, you need to ensure that you _understand_ where the bottlenecks are, so that you know which parts of the code _should_ be re-written.
One way in which we can find those bottlenecks is through _profiling_ - running our code, and timing each part of it, to identify which parts are causing our program to run slowly.
As a simple example, consider this code. It is running slowly, and we want to know why:
%% Cell type:code id:92119f45 tags:
```
import numpy as np
def process_data(datafile):
data = np.loadtxt(datafile)
result = data.mean()
return result
print(process_data('mydata.txt'))
```
%% Cell type:markdown id:28a3e8e1 tags:
> The `mydata.txt` file can be generated by running this code:
>
> ```
> import numpy as np
> data = np.random.random((10000, 1000))
> np.savetxt('mydata.txt', data)
> ```
For small functions, you can profile code manually by using the `time` module, e.g.:
%% Cell type:code id:49fa2ba9 tags:
```
import time
start = time.time()
result = process_data('mydata.txt')
end = time.time()
print(f'Total #seconds: {end - start}')
print(result)
```
%% Cell type:markdown id:6c6de92c tags:
We could also do the same within the `process_data` function, and calculate and report the timings for each individual section within. But there is a library called `line_profiler` which can do this for you - it will time every single line of code, and print a report for you:
%% Cell type:code id:2413b91c tags:
```
%load_ext line_profiler
%lprun -f process_data process_data('mydata.txt')
```
%% Cell type:markdown id:d3cf2572 tags:
> `line_profiler` is not installed as part of FSL by default, but you can install it with this command:
> ```$FSLDIR/bin/conda install -p $FSLDIR line_profiler```
>
> And you can also use it from the command-line, if you are not working in a Jupyter notebook. You can learn more about `line_profiler` at https://github.com/pyutils/line_profiler
We can see that most of the processing time was actually in _loading_ the data from file, and not in the data _processing_ (simply calculating the mean in this toy example). Without profiling your code, you may have naively thought that the data processing step was the culprit, and needed to be optimised. But what profiling has told us here is that our problem lies elsewhere, and perhaps we need to think about changing how we are storing our data.
But let's assume from this point on that we _do_ need to optimise our code...
<a class="anchor" id="joblib"></a>
# JobLib
https://joblib.readthedocs.io/en/stable/
JobLib has been around for a while, and is a simple and lightweight library with two main features:
- An API for "embarrassingly parallel" tasks.
- An API for caching/re-using the results of time-consuming computations.
- An API for caching/re-using the results of time-consuming computations (which we won't discuss in this notebook, but you can read about [here](https://joblib.readthedocs.io/en/stable/memory.html)).
On the surface, JobLib does not provide any functionality that cannot already be accomplished with built-in libraries such as `multiprocessing` and `functools`. However, the JobLib developers have put a lot of effort into ensuring that JobLib:
- is easy to use,
- works consistently across all of the major platforms, and
- works as efficiently as possible with Numpy arrays
We can use the JobLib API by importing the `joblib` package (we'll also use `numpy` and `time` in this example):
%% Cell type:code id:5791f029 tags:
%% Cell type:code id:0198ff8d tags:
```
import numpy as np
import joblib
import time
```
%% Cell type:markdown id:ecf0ad57 tags:
%% Cell type:markdown id:f64fac25 tags:
Now, let's say that we have a collection of `numpy` arrays containing image data for a group of subjects:
%% Cell type:code id:53839192 tags:
%% Cell type:code id:a79aacb0 tags:
```
datasets = [np.random.random((91, 109, 91)) for i in range(10)]
```
%% Cell type:markdown id:733f66cc tags:
%% Cell type:markdown id:1e30049b tags:
And we want to calculate some metric on each image. We simply need to define a function which contains the calculation we want to perform (the `time.sleep` call is there just to simulate a complex calculation):
%% Cell type:code id:049cd8dd tags:
%% Cell type:code id:ba63be7a tags:
```
def do_calculation(input_data):
time.sleep(2)
return input_data.mean()
```
%% Cell type:markdown id:1266287d tags:
%% Cell type:markdown id:d3d2b0b8 tags:
We can now use `joblib.Parallel` and `joblib.delayed` to parallelise those calculations. `joblib.Parallel` sets up a pool of worker processes, and `joblib.delayed` schedules a single instantiation of the function, and associated data, which is to be executed. We can execute a sequence of `delayed` tasks by passing them to the `Parallel` pool:
%% Cell type:code id:931cb5d7 tags:
%% Cell type:code id:814aef5c tags:
```
with joblib.Parallel(n_jobs=-1) as pool:
tasks = [joblib.delayed(do_calculation)(d) for d in datasets]
results = pool(tasks)
print(results)
```
%% Cell type:markdown id:5b499444 tags:
%% Cell type:markdown id:e3bd3114 tags:
Just like with `multiprocessing`, JobLib is susceptible to the same problems with regard to sharing data between processes (these problems are
discussed in the `advanced_programming/threading.ipynb` notebook). In the example above, each dataset is serialised and copied from the main process
to the worker processes, and then the result copied back. This behaviour could be a performance bottleneck for your own task, or you may be working
on a system with limited memory which is incapable of storing several copies of the data.
Just like with `multiprocessing`, JobLib is susceptible to the same problems with regard to sharing data between processes (these problems are discussed in the `advanced_programming/threading.ipynb` notebook). In the example above, each dataset is serialised and copied from the main process to the worker processes, and then the result copied back. This behaviour could be a performance bottleneck for your own task, or you may be working on a system with limited memory which is incapable of storing several copies of the data.
To deal with this, we can use memory-mapped Numpy arrays. This is a feature built into Numpy, and supported by JobLib, which stores your data in your file system instead of in memory. This allows your data to be simultaneously read and written by multiple processes.
To deal with this, we can use memory-mapped Numpy arrays. This is a feature built into Numpy, and supported by JobLib, which invlolves storing your data in your file system instead of in memory. This allows your data to be simultaneously read and written by multiple processes.
> Some other options for sharing data between processes are discussed in the `advanced_programming/threading.ipynb` notebook.
You can create `numpy.memmap` arrays directly, but JobLib will also automatically convert any Numpy arrays which are larger than a specified threshold into memory-mapped arrays. This threshold defaults to 1MB, and you can change it via the `n_bytes` option when you create a `joblib.Parallel` pool.
To see this in action, imagine that you have some 4D fMRI data, and wish to fit a complicated model to the time series at each voxel. We will write our model fitting function so that it works on one slice at a time:
%% Cell type:code id:6ec1fb3a tags:
%% Cell type:code id:9b7106d6 tags:
```
def fit_model(indata, outdata, sliceidx):
print(f'Fitting model at slice {sliceidx}')
time.sleep(1)
outdata[:, :, sliceidx] = indata[:, :, sliceidx, :].mean() + sliceidx
```
%% Cell type:markdown id:314c22e5 tags:
%% Cell type:markdown id:44eff256 tags:
Now we will load our 4D data, and pre-allocate another array to store the fitted model parameters.
%% Cell type:code id:d9f90609 tags:
%% Cell type:code id:2e548f25 tags:
```
# Imagine that we have loaded this data from a file
data = np.random.random((91, 109, 91, 50)).astype(np.float32)
# Pre-allocate space to store the fitted model parameters
model = np.memmap('model.mmap',
shape=(91, 109, 91),
dtype=np.float32,
mode='w+')
# Fit our model, processing slices in parallel
with joblib.Parallel(n_jobs=-1) as pool:
pool(joblib.delayed(fit_model)(data, model, slc) for slc in range(91))
print(model)
```
%% Cell type:markdown id:47dc9a3b tags:
%% Cell type:markdown id:c5a776e2 tags:
<a class="anchor" id="dask"></a>
# Dask
https://www.dask.org/
Dask is a very powerful library which you can use to parallelise your code, and to distribute computation across multiple machines. You can use Dask locally on your own computer, on HPC clusters (e.g. SLURM and SGE) and with cloud compute platforms (e.g. AWS, Azure).
Dask has two main components:
- APIs for defining tasks/jobs - there is a low-level API similar to that provided by JobLib, but Dask also has sophisticated high-level APIs for working with Pandas and Numpy-style data.
- APIs for defining tasks/jobs - there is a low-level API comparable to that provided by JobLib, but Dask also has sophisticated high-level APIs for working with Pandas and Numpy-style data.
- A task scheduler which builds a graph of all the tasks that need to be executed, and which manges their execution, either locally or remotely.
We will introduce the Numpy API and will briefly cover the low-level API, and then demonstrate how to use Dask to perform calculations on a SGE cluster.
We will introduce the Numpy API and the low-level API, and then demonstrate how to use Dask to perform calculations on a SGE cluster.
<a class="anchor" id="dask-numpy-api"></a>
## Dask Numpy API
https://docs.dask.org/en/stable/array.html
> Dask also provides a Pandas-style API, accessible in the `dask.dataframe` package - you can read about it at https://docs.dask.org/en/stable/dataframe.html.
To use the Dask Numpy API, simply import the `dask.array` package, and use it instead of the `numpy` package:
%% Cell type:code id:7923a7be tags:
%% Cell type:code id:10dabd47 tags:
```
import numpy as np
import dask.array as da
data = da.random.random((1000, 1000, 1000, 20)).astype(np.float32)
data
```
%% Cell type:markdown id:2bdd29e6 tags:
%% Cell type:markdown id:6c7c7ecd tags:
If you do the numbers, you will realise that the above call has created an array which requires **74 gigabytes** of memory - this is far more memory than what is available in most consumer level computers.
> The call would almost certainly fail if you made it using `np.random.random` instead of `da.random.random`, as Numpy will attempt to create the entire data set (and in fact would temporarily require up to **150 gigabytes** of memory, as Numpy uses double-precision floating point (`float64`) values by default).
The call would almost certainly fail if you made it using `np.random.random` instead of `da.random.random`, as Numpy would attempt to create the entire data set (and in fact would temporarily require up to **150 gigabytes** of memory, as Numpy uses double-precision floating point (`float64`) values by default).
However:
However, this worked because:
- Dask automatically splits arrays into smaller chunks - in the above code, `data` has been split into 1331 smaller arrays, each of which has shape `(94, 94, 94, 10)`, and requires only 64 megabytes.
- Most operations in the `dask.array` package are _lazily evaluated_. The above call has not actually created any data - Dask has just created a set of jobs which describe how to create a large array of random values.
When using `dask.array`, nothing is actually executed until you call the `compute()` function. For example, we can try calculating the mean of our large array:
%% Cell type:code id:b2605c92 tags:
%% Cell type:code id:724334d6 tags:
```
m = data.mean()
m
```
%% Cell type:markdown id:08bf1c77 tags:
%% Cell type:markdown id:98438c67 tags:
But again, this will not actually calculate the mean - it just defines a task which, when executed, will calculate the mean over the `data` array.
We can _execute_ that task by callintg `compute()` on the result:
%% Cell type:code id:cecf0b61 tags:
%% Cell type:code id:fe23c705 tags:
```
print(m.compute())
```
%% Cell type:markdown id:1472b53d tags:
%% Cell type:markdown id:e558f8cb tags:
Dask arrays support _most_ of the functionality of Numpy arrays, but there are a few exceptions, most notably the `numpy.linalg` package.
Dask arrays support most of the functionality of Numpy arrays, but there are a few exceptions, most notably the `numpy.linalg` package.
> Remember that Dask also provides a Pandas-style API, accessible in the `dask.dataframe` package - you can read about it at https://docs.dask.org/en/stable/dataframe.html.
<a class="anchor" id="low-level-dask-api"></a>
## Low-level Dask API
In addition to the Numpy and Pandas APIs, Dask also has a lower-level interface which allows you to create and lazily execute a pipeline made up of Python functions. This API is based around `dask.delayed`, which is a Python decorator that can be applied to any function, and which tells Dask that a function should be lazily executed.
As a very simple example (taken from the [Dask documentation](https://docs.dask.org/en/stable/delayed.html#example)), consider this simple numerical task, where we have a set of numbers and, for each number `x`, we want to calculate `(x * x)`, and then add all of the results together (i.e. the sum of squares). We'll start by defining a function for each of the operations that we need to perform:
%% Cell type:code id:1f764830 tags:
%% Cell type:code id:9c3acf29 tags:
```
def square(x):
return x * x
def sum(values):
total = 0
for v in values:
total = total + v
return total
```
%% Cell type:markdown id:fd5252f6 tags:
%% Cell type:markdown id:09c6f537 tags:
We could solve this problem without parallelism by using conventional Python code, which might look something like the following:
%% Cell type:code id:b0910aaf tags:
%% Cell type:code id:77e896a4 tags:
```
data = [1, 2, 3, 4, 5]
output = []
for x in data:
s = square(x)
output.append(s)
total = sum(output)
print(total)
```
%% Cell type:markdown id:8a7a45f5 tags:
%% Cell type:markdown id:b731893c tags:
However, this problem is inherently parallelisable - we are independently performing the same series of steps to each of our inputs, before the final tallying. We can use the `dask.delayed` function to take advantage of this:
%% Cell type:code id:9031f120 tags:
%% Cell type:code id:a79b728e tags:
```
import dask
output = []
for x in data:
a = dask.delayed(square)(x)
output.append(a)
total = dask.delayed(sum)(output)
```
%% Cell type:markdown id:6bc3ef4c tags:
%% Cell type:markdown id:1cccc09a tags:
We have not actually performed any computations yet. What we have done is built a _graph_ of operations that we need to perform. Dask refers to this as a **task graph** - Dask keeps track of the dependencies of each function that we add, and so is able to determine the order in they need to be executed, and also which steps do not depend on each other and so can be executed in parallel. Dask can even generate an image of our task graph for us:
We have not actually performed any computations yet. What we have done is built a _graph_ of operations that we need to perform. Dask refers to this as a **task graph**. Dask keeps track of the dependencies of each function that we add, and so is able to determine the order in they need to be executed, and also which steps do not depend on each other and so can be executed in parallel. Dask can even visualise this task graph for us:
%% Cell type:code id:9f0833d9 tags:
%% Cell type:code id:8d9b95b0 tags:
```
total.visualize()
```
%% Cell type:markdown id:0164c407 tags:
%% Cell type:markdown id:9c387b05 tags:
And then when we are ready, we can call `compute()` to actually do the calculation:
%% Cell type:code id:8561997b tags:
%% Cell type:code id:f5d1d09c tags:
```
total.compute()
```
%% Cell type:markdown id:3590a9f9 tags:
%% Cell type:markdown id:c1c95d5f tags:
For a more realistic example, let's imagine that we have T1 MRI images for five subjects, and we want to perform basic structural preprocessing on each of them (reorientation, FOV reduction, and brain extraction). He're creating this example data set (all of the T1 images are just a copy of the `bighead.nii.gz` image, from the FSL course data).
For a more realistic example, let's imagine that we have T1 MRI images for five subjects, and we want to perform basic structural preprocessing on each of them (reorientation, FOV reduction, and brain extraction). Here we're creating this example data set (all of the T1 images are just a copy of the `bighead.nii.gz` image, from the FSL course data).
%% Cell type:code id:c3879db5 tags:
%% Cell type:code id:f6f614f2 tags:
```
import os
import shutil
os.makedirs('braindata', exist_ok=True)
for i in range(1, 6):
shutil.copy('../../applications/fslpy/bighead.nii.gz', f'braindata/{i:02d}.nii.gz')
```
%% Cell type:markdown id:77851c91 tags:
%% Cell type:markdown id:ababc17d tags:
And now we can build our pipeline. The [fslpy](https://open.win.ox.ac.uk/pages/fsl/fslpy/) library has a collection of functions which we can use to call the FSL commands that we need.
And now we can build our pipeline. The [fslpy](https://open.win.ox.ac.uk/pages/fsl/fslpy/) library has a collection of functions which we can use to call the FSL commands that we need. We need to do a little work, as by default `dask.delayed` assumes that the dependencies of a task are passed in as arguments (there are other ways of dealing with this, but in this example it is easy to simply write a few small functions that define each of our tasks):
%% Cell type:code id:1f9d5f36 tags:
We need to do a little work, as by default `dask.delayed` assumes that the dependencies of a function (i.e. other `delayed` functions) are passed to the function as arguments. There are [other methods](https://docs.dask.org/en/stable/delayed.html#indirect-dependencies) of dealing with this, but often the easiest option is simply to write a few small functions that define each of our tasks, in a manner that satisfies this assumption:
%% Cell type:code id:fe917598 tags:
```
import fsl.wrappers as fw
def reorient(input, output):
fw.fslreorient2std(input, output)
return output
def fov(input, output):
fw.robustfov(input, output)
return output
def bet(input, output):
fw.bet(input, output)
return output
```
%% Cell type:code id:4a762719 tags:
%% Cell type:markdown id:a1cd4b06 tags:
Again we use `dask.delayed` to build up a graph of tasks that need executing, starting from the input files, and ending with our final brain-extracted outputs. Again, we are not actually executing anything here - we're just building up a task graph that Dask will then execute for us later:
%% Cell type:code id:d03c2ae2 tags:
```
import glob
import dask
import fsl.data.image as fslimage
inputs = list(glob.glob('braindata/??.nii.gz'))
tasks = []
for input in inputs:
basename = fslimage.removeExt(input)
r = dask.delayed(reorient)(input, f'{basename}_reorient.nii.gz')
f = dask.delayed(fov)(r, f'{basename}_fov.nii.gz')
b = dask.delayed(bet)(f, f'{basename}_brain.nii.gz')
tasks.append(b)
```
%% Cell type:markdown id:1ed04761 tags:
%% Cell type:markdown id:e0215eaf tags:
In the previous example we had a single output (the result of summing the squared input values) which we called `visualize()` on. We can also call `dask.visualize()` to visualise a collection of tasks:
In the previous example we had a single output (the result of summing the squared input values) upon which we called `visualize()`. Here we have a list of independent tasks (in the `tasks` list). However, we can still visualise them by passing the entire list to the `dask.visualize()` function:
%% Cell type:code id:f62c7c6d tags:
%% Cell type:code id:a08d0d91 tags:
```
dask.visualize(*tasks)
```
%% Cell type:markdown id:63017c42 tags:
%% Cell type:markdown id:244a2ae9 tags:
And, similarly, we can call `dask.compute` to run them all:
And, similarly, we can call `dask.compute` to run them all at once:
%% Cell type:code id:7b6e40a9 tags:
%% Cell type:code id:6b98bb2a tags:
```
outputs = dask.compute(*tasks)
print(outputs)
```
%% Cell type:markdown id:5dbb82f6 tags:
%% Cell type:markdown id:6d77a9f8 tags:
<a class="anchor" id="distributing-computation-with-dask"></a>
## Distributing computation with Dask
In the examples above, Dask was running its tasks in parallel on your local machine. But it is easy to instruct Dask to distribute its computations across multiple machines. For example, Dask has support for executing tasks on a SGE or SLURM cluster, which we have available in Oxford at FMRIB and the BDI.
In the examples above, Dask was running your tasks in parallel on your local machine. But it is easy to instruct Dask to distribute your computations across multiple machines. For example, Dask has support for executing tasks on a SGE or SLURM cluster, which we have available in Oxford at FMRIB and the BDI.
To use this functionality, we need an additional library called `dask-jobqueue` which, at the moment, is not installed as part of FSL.
> Note that the code cells below will not work if you are running this notebook on your own computer (unless you happen to have a SGE cluster system installed on your laptop).
The first step is to create a `SGECluster` (or `SLURMCluster`) object. You need to populate this object with information about a _single job_ in your workflow. At a minumum, you must specify the number of cores and memory that are available to a single job:
%% Cell type:code id:b6e7814e tags:
%% Cell type:code id:e206a399 tags:
```
from dask_jobqueue import SGECluster
cluster = SGECluster(cores=2, memory='16GB')
```
%% Cell type:markdown id:10dd646c tags:
%% Cell type:markdown id:a69e8eaa tags:
You can also set a range of other options, including specifying a queue name (e.g. `queue='short.q'`), or specifying the total amount of time that your job will run for (e.g. `walltime='01:00:00'` for one hour).
The next step is to create some jobs by calling the `scale()` method:
%% Cell type:code id:bfa74705 tags:
%% Cell type:code id:20c0da1a tags:
```
cluster.scale(jobs=5)
```
%% Cell type:markdown id:7123a281 tags:
%% Cell type:markdown id:96c05bcd tags:
Behind the scenes, the `scale()` method uses `qsub` to create five "worker processes", each running on a different cluster node. In the first instance, these worker processes will be sitting idle, doing nothing, and waiting for you to give them a task to run.
The final step is to create "client" object. This is an important step which configures Dask to use the cluster we have just set up:
%% Cell type:code id:0c8502aa tags:
%% Cell type:code id:e3742ee0 tags:
```
client = cluster.get_client()
```
%% Cell type:markdown id:9b43cee7 tags:
%% Cell type:markdown id:d4053813 tags:
After creating a client, any Dask computations that we request will be performed on the cluster by our worker processes:
%% Cell type:code id:c06367ee tags:
%% Cell type:code id:2b8fc2c0 tags:
```
import dask.array as da
data = da.random.random((1000, 1000, 1000, 10))
print(data.mean().compute())
```
%% Cell type:markdown id:f5c78ed3 tags:
%% Cell type:markdown id:e0e99795 tags:
<a class="anchor" id="fsl-pipe"></a>
## fsl-pipe
# fsl-pipe
https://open.win.ox.ac.uk/pages/fsl/fsl-pipe/
fsl-pipe is a Python library (written by our very own Michiel Cottaar) which builds upon another library called [file-tree](https://open.win.ox.ac.uk/pages/fsl/file-tree/), and allows you to write an analysis pipeline in a _declarative_ manner. A pipeline is defined by:
- A "file tree" which defines the directory structure of the input and output files of the pipeline.
- A set of recipes (Python functions) describing how each of the pipeline outputs should be produced.
- A _file tree_ which defines the directory structure of the input and output files of the pipeline.
- A set of _recipes_ (Python functions) describing how each of the pipeline outputs should be produced.
In a similar vein to Dask task graphs, fsl-pipe will automatically determine which recipes should be executed, and in what order, to produce whichever output file(s) you request. Recipes can either run locally, be distributed using Dask, or be executed on a cluster using `fsl_sub`.
In a similar vein to [Dask task graphs](#low-level-dask-api), fsl-pipe will automatically determine which recipes should be executed, and in what order, to produce whichever output file(s) you request. You can instruct fsl-pipe to run your recipes locally, be parallelised or distributed using Dask, or be executed on a cluster using `fsl_sub`.
For example, let's again imagine that we have some T1 images upon which we wish to perform basic structural preprocessing, and which arranged in the following manner:
For example, let's again imagine that we have some T1 images upon which we wish to perform basic structural preprocessing, and which are arranged in the following manner:
> ```
> subjectA/
> T1w.nii.gz
> subjectB/
> T1w.nii.gz
> subjectC/
> T1w.nii.gz
> ```
The code cell below will automatically create a dummy data set with the above structure:
%% Cell type:code id:4d42fa32 tags:
```
import os
import shutil
for subj in 'ABC':
subjdir = f'mydata/subject{subj}'
os.makedirs(subjdir, exist_ok=True)
shutil.copy('../../applications/fslpy/bighead.nii.gz', f'{subjdir}/T1w.nii.gz')
```
%% Cell type:markdown id:01575655 tags:
We must first describe the structure of our data, and save that description as a file-tree file (e.g. `mydata.tree`). This file contains a placeholder for the subject ID, and gives both the input and any desired output files unique identifiers:
%% Cell type:code id:4de54ba7 tags:
```
%%writefile mydata.tree
subject{subject}
T1w.nii.gz (t1)
T1w_brain.nii.gz (t1_brain)
T1w_fov.nii.gz (t1_fov)
T1w_reorient.nii.gz (t1_reorient)
```
%% Cell type:markdown id:99006f8c tags:
Now we need to define our _recipes_, the individual processing steps that we want to perform, and that will generate our output files. This is similar to what we did above when we were using [Dask](#low-level-dask-api) - we define functions which perform each step.
%% Cell type:code id:819900fa tags:
```
from fsl.wrappers import bet, fslreorient2std, robustfov
from fsl_pipe import Pipeline, In, Out
def reorient(t1 : In, t1_reorient : Out):
fslreorient2std(t1, t1_reorient)
def fov(t1_reorient : In, t1_fov : Out):
robustfov(t1_reorient, t1_fov)
def brain_extract(t1_fov : In, t1_brain : Out):
bet(t1_fov, t1_brain)
```
%% Cell type:markdown id:91375ad3 tags:
Note that we have also annotated the inputs and outputs of each function, and have used the same identifiers that we used in our `mydata.tree` file above. By doing this, `fsl-pipe` is automatically able to determine the steps that would be required in order to generate the output files that we request.
Once we have defined all of our recipes, we need to create a `Pipeline` object, and add all of our functions to it:
%% Cell type:code id:25e48fd7 tags:
```
pipe = Pipeline()
pipe(reorient)
pipe(fov)
pipe(brain_extract)
```
%% Cell type:markdown id:d1d5024f tags:
> Note that it is also possible (and equivalent) to create the `Pipeline` before defining our recipe functions, and to use the pipeline as a decorator, e.g.:
>
> ```
> subject{subject}
> T1w.nii.gz (t1)
> T1w_brain.nii.gz (t1_brain)
> T1w_fov.nii.gz (t1_fov)
> T1w_reorient.nii.gz (t1_reorient)
> pipe = Pipeline()
>
> @pipe
> def reorient(t1 : In, t1_reorient : Out):
> ...
> ```
>
We now need to create a `FileTree` object which is used to generate file paths for the input and output files. The `update_glob('t1')` method instructs the `FileTree` to scan the file system, and to generate file paths for all T1 images that are present.
%% Cell type:code id:9d49246c tags:
```
from file_tree import FileTree
tree = FileTree.read('mydata.tree', './mydata/').update_glob('t1')
```
%% Cell type:markdown id:84d9df1f tags:
%% Cell type:code id:07ff84b0 tags:
Then it is very easy to run our pipeline:
%% Cell type:code id:d0c92b06 tags:
```
#!/usr/bin/env python
jobs = pipe.generate_jobs(tree)
jobs.run()
```
%% Cell type:markdown id:e486a07f tags:
The default behaviour when using fsl-pipe locally is to for one task to be executed at a time. However, if you run fsl-pipe on the cluster, it will automatically submit the jobs using `fsl_sub`. You can also tell fsl-pipe to execute the pipeline using Dask, which will cause any independent jobs to be executed in parallel, e.g.:
%% Cell type:code id:a61d7603 tags:
```
jobs.run(method='dask')
```
%% Cell type:markdown id:458be3fe tags:
The above example is just one way of using fsl-pipe - the library has several powerful features, including its own command-line interface, and the ability to skip jobs for output files that already exist.
<a class="anchor" id="fsl-sub"></a>
## fsl-sub
You can use the venerable `fsl_sub` to run several tasks in parallel both on your local machine, and on a HPC cluster, such as the ones available to us at FMRIB and the BDI. `fsl_sub` is typically called from the command-line, e.g.:
> ```
> for dataset in mydata/sub-*; do
> fsl_sub ./my_processing_script.py ${dataset} --jobram 16
> done
> ```
If you are working on a local machine, `fsl_sub` will block until your command has completed. You can run multiple commands in parallel by using "array tasks" - save each of the commands you want to run to a text file, e.g.:
> ```
> echo "./my_processing_script.py mydata/sub-01" > tasks.txt
> echo "./my_processing_script.py mydata/sub-02" >> tasks.txt
> echo "./my_processing_script.py mydata/sub-03" >> tasks.txt
> echo "./my_processing_script.py mydata/sub-04" >> tasks.txt
> echo "./my_processing_script.py mydata/sub-05" >> tasks.txt
> ```
And then run `fsl_sub -t tasks.txt` - each of your commands will be executed in parallel. If you are working on a cluster, `fsl_sub` will schedule all of the commands to be executed simultaneously.
You can also call `fsl_sub` from Python by using functions from the `fslpy` library:
> ```
> from glob import glob
> from fsl.wrappers import fsl_sub
>
> for dataset in glob('mydata/sub-*'):
> fsl_sub(f'./my_processing_script.py ${dataset}', jobram=16)
> ```
And the `fslpy` wrapper functions allow you to run FSL commands with `fsl_sub`, and to specify job dependencies. For example, to submit a collection of jobs, you can pass `submit=True`:
%% Cell type:code id:37134ca6 tags:
```
from fsl.data.image import removeExt
from fsl.wrappers import bet
from glob import glob
from file_tree import FileTree
from fsl_pipe import pipe, In, Out
for t1 in glob('braindata/??.nii.gz'):
t1 = removeExt(t1)
bet(t1, f'{t1}_brain', submit={'jobram':16})
```
@pipe
def brain_extract(t1 : In, bet_output : Out, bet_mask : Out):
bet(t1, bet_output, mask=True)
%% Cell type:markdown id:681b1a8c tags:
You can also specify dependencies by using the ID of a previously submitted job, e.g.:
%% Cell type:code id:e7c2329c tags:
```
from fsl.data.image import removeExt
from fsl.wrappers import robustfov, bet
from glob import glob
tree = FileTree.read('data.tree').update_glob('t1')
pipe.cli(tree)
for t1 in glob('braindata/??.nii.gz'):
t1 = removeExt(t1)
jid = robustfov(t1, f'{t1}_fov', submit=True)
bet(f'{t1}_fov', f'{t1}_brain', submit={'jobram':16, 'jobhold' : jid})
```
......
applications/parallel/parallel.md
+ 260
− 41
View file @ 0e86b1fc
......@@ -4,16 +4,104 @@
While Python has built-in support for threading and parallelising in the
[`threading`](https://docs.python.org/3/library/threading.html) and
[`multiprocessing`](https://docs.python.org/3/library/multiprocessing.html)
modules, there are a range of third-party libraries which you can use to improve the performance of your code.
modules (covered in `advanced_programming/threading.ipynb`), there are a range of third-party libraries which you can use to improve the performance of your code.
Contents:
* [Do you really need to parallelise your code?](#do-you-need-to-parallelise)
* [Profilng your code](#profiling-your-code)
* [JobLib](#joblib)
* [Dask](#dask)
* [Dask Numpy API](#dask-numpy-api)
* [Low-level Dask API](#low-level-dask-api)
* [Distributing computation with Dask](#distributing-computation-with-dask)
* [fsl-pipe](#fsl-pipe)
* [fsl-sub](#fsl-sub)
<a class="anchor" id="do-you-need-to-parallelise"></a>
# Do you really need to parallelise your code?
Before diving in and tearing your code apart, you should think very carefully about your problem, and the hardware you are using to run your code. For example, if you are processing MRI data for a number of subjects on the FMRIB cluster (where each node on the cluster has fairly modest hardware specs, meaning that within-process parallelisation will have limited benefits), the most efficient option may be to write your code in a single-threaded manner, and to parallelise across data sets, processing one data set on each cluster node.
Your best option may simply be to repeatedly call `fsl_sub`- see the [section below](#fsl-sub) for more details on this approach.
<a class="anchor" id="#profiling-your-code"></a>
## Profilng your code
Once you have decided that you need to parallelise some steps in your program, **STOP!** As the great Donald Knuth once said:
> Programmers waste enormous amounts of time thinking about, or worrying about, the speed of noncritical parts of their programs, and these attempts at efficiency actually have a strong negative impact when debugging and maintenance are considered. We should forget about small efficiencies, say about 97% of the time:
>
> **premature optimization is the root of all evil**.
>
> Yet we should not pass up our opportunities in that critical 3%.
What Knuth is essentially saying is that there is no point in optimising or rewriting a piece of code unless it is actually going to have a positive impact on performance. In other words, before you refactor your code, you need to ensure that you _understand_ where the bottlenecks are, so that you know which parts of the code _should_ be re-written.
One way in which we can find those bottlenecks is through _profiling_ - running our code, and timing each part of it, to identify which parts are causing our program to run slowly.
As a simple example, consider this code. It is running slowly, and we want to know why:
```
import numpy as np
def process_data(datafile):
data = np.loadtxt(datafile)
result = data.mean()
return result
print(process_data('mydata.txt'))
```
> The `mydata.txt` file can be generated by running this code:
>
> ```
> import numpy as np
> data = np.random.random((10000, 1000))
> np.savetxt('mydata.txt', data)
> ```
For small functions, you can profile code manually by using the `time` module, e.g.:
```
import time
start = time.time()
result = process_data('mydata.txt')
end = time.time()
print(f'Total #seconds: {end - start}')
print(result)
```
We could also do the same within the `process_data` function, and calculate and report the timings for each individual section within. But there is a library called `line_profiler` which can do this for you - it will time every single line of code, and print a report for you:
```
%load_ext line_profiler
%lprun -f process_data process_data('mydata.txt')
```
> `line_profiler` is not installed as part of FSL by default, but you can install it with this command:
> ```$FSLDIR/bin/conda install -p $FSLDIR line_profiler```
>
> And you can also use it from the command-line, if you are not working in a Jupyter notebook. You can learn more about `line_profiler` at https://github.com/pyutils/line_profiler
We can see that most of the processing time was actually in _loading_ the data from file, and not in the data _processing_ (simply calculating the mean in this toy example). Without profiling your code, you may have naively thought that the data processing step was the culprit, and needed to be optimised. But what profiling has told us here is that our problem lies elsewhere, and perhaps we need to think about changing how we are storing our data.
But let's assume from this point on that we _do_ need to optimise our code...
<a class="anchor" id="joblib"></a>
......@@ -25,7 +113,7 @@ https://joblib.readthedocs.io/en/stable/
JobLib has been around for a while, and is a simple and lightweight library with two main features:
- An API for "embarrassingly parallel" tasks.
- An API for caching/re-using the results of time-consuming computations.
- An API for caching/re-using the results of time-consuming computations (which we won't discuss in this notebook, but you can read about [here](https://joblib.readthedocs.io/en/stable/memory.html)).
On the surface, JobLib does not provide any functionality that cannot already be accomplished with built-in libraries such as `multiprocessing` and `functools`. However, the JobLib developers have put a lot of effort into ensuring that JobLib:
......@@ -66,12 +154,11 @@ with joblib.Parallel(n_jobs=-1) as pool:
print(results)
```
Just like with `multiprocessing`, JobLib is susceptible to the same problems with regard to sharing data between processes (these problems are
discussed in the `advanced_programming/threading.ipynb` notebook). In the example above, each dataset is serialised and copied from the main process
to the worker processes, and then the result copied back. This behaviour could be a performance bottleneck for your own task, or you may be working
on a system with limited memory which is incapable of storing several copies of the data.
Just like with `multiprocessing`, JobLib is susceptible to the same problems with regard to sharing data between processes (these problems are discussed in the `advanced_programming/threading.ipynb` notebook). In the example above, each dataset is serialised and copied from the main process to the worker processes, and then the result copied back. This behaviour could be a performance bottleneck for your own task, or you may be working on a system with limited memory which is incapable of storing several copies of the data.
To deal with this, we can use memory-mapped Numpy arrays. This is a feature built into Numpy, and supported by JobLib, which stores your data in your file system instead of in memory. This allows your data to be simultaneously read and written by multiple processes.
To deal with this, we can use memory-mapped Numpy arrays. This is a feature built into Numpy, and supported by JobLib, which invlolves storing your data in your file system instead of in memory. This allows your data to be simultaneously read and written by multiple processes.
> Some other options for sharing data between processes are discussed in the `advanced_programming/threading.ipynb` notebook.
You can create `numpy.memmap` arrays directly, but JobLib will also automatically convert any Numpy arrays which are larger than a specified threshold into memory-mapped arrays. This threshold defaults to 1MB, and you can change it via the `n_bytes` option when you create a `joblib.Parallel` pool.
......@@ -115,23 +202,19 @@ Dask is a very powerful library which you can use to parallelise your code, and
Dask has two main components:
- APIs for defining tasks/jobs - there is a low-level API similar to that provided by JobLib, but Dask also has sophisticated high-level APIs for working with Pandas and Numpy-style data.
- APIs for defining tasks/jobs - there is a low-level API comparable to that provided by JobLib, but Dask also has sophisticated high-level APIs for working with Pandas and Numpy-style data.
- A task scheduler which builds a graph of all the tasks that need to be executed, and which manges their execution, either locally or remotely.
We will introduce the Numpy API and will briefly cover the low-level API, and then demonstrate how to use Dask to perform calculations on a SGE cluster.
We will introduce the Numpy API and the low-level API, and then demonstrate how to use Dask to perform calculations on a SGE cluster.
<a class="anchor" id="dask-numpy-api"></a>
## Dask Numpy API
https://docs.dask.org/en/stable/array.html
> Dask also provides a Pandas-style API, accessible in the `dask.dataframe` package - you can read about it at https://docs.dask.org/en/stable/dataframe.html.
To use the Dask Numpy API, simply import the `dask.array` package, and use it instead of the `numpy` package:
```
......@@ -144,10 +227,9 @@ data
If you do the numbers, you will realise that the above call has created an array which requires **74 gigabytes** of memory - this is far more memory than what is available in most consumer level computers.
> The call would almost certainly fail if you made it using `np.random.random` instead of `da.random.random`, as Numpy will attempt to create the entire data set (and in fact would temporarily require up to **150 gigabytes** of memory, as Numpy uses double-precision floating point (`float64`) values by default).
The call would almost certainly fail if you made it using `np.random.random` instead of `da.random.random`, as Numpy would attempt to create the entire data set (and in fact would temporarily require up to **150 gigabytes** of memory, as Numpy uses double-precision floating point (`float64`) values by default).
However:
However, this worked because:
- Dask automatically splits arrays into smaller chunks - in the above code, `data` has been split into 1331 smaller arrays, each of which has shape `(94, 94, 94, 10)`, and requires only 64 megabytes.
......@@ -170,7 +252,10 @@ print(m.compute())
```
Dask arrays support most of the functionality of Numpy arrays, but there are a few exceptions, most notably the `numpy.linalg` package.
Dask arrays support _most_ of the functionality of Numpy arrays, but there are a few exceptions, most notably the `numpy.linalg` package.
> Remember that Dask also provides a Pandas-style API, accessible in the `dask.dataframe` package - you can read about it at https://docs.dask.org/en/stable/dataframe.html.
<a class="anchor" id="low-level-dask-api"></a>
......@@ -220,7 +305,7 @@ for x in data:
total = dask.delayed(sum)(output)
```
We have not actually performed any computations yet. What we have done is built a _graph_ of operations that we need to perform. Dask refers to this as a **task graph** - Dask keeps track of the dependencies of each function that we add, and so is able to determine the order in they need to be executed, and also which steps do not depend on each other and so can be executed in parallel. Dask can even generate an image of our task graph for us:
We have not actually performed any computations yet. What we have done is built a _graph_ of operations that we need to perform. Dask refers to this as a **task graph**. Dask keeps track of the dependencies of each function that we add, and so is able to determine the order in they need to be executed, and also which steps do not depend on each other and so can be executed in parallel. Dask can even visualise this task graph for us:
```
total.visualize()
......@@ -233,7 +318,7 @@ total.compute()
```
For a more realistic example, let's imagine that we have T1 MRI images for five subjects, and we want to perform basic structural preprocessing on each of them (reorientation, FOV reduction, and brain extraction). He're creating this example data set (all of the T1 images are just a copy of the `bighead.nii.gz` image, from the FSL course data).
For a more realistic example, let's imagine that we have T1 MRI images for five subjects, and we want to perform basic structural preprocessing on each of them (reorientation, FOV reduction, and brain extraction). Here we're creating this example data set (all of the T1 images are just a copy of the `bighead.nii.gz` image, from the FSL course data).
```
......@@ -245,7 +330,10 @@ for i in range(1, 6):
shutil.copy('../../applications/fslpy/bighead.nii.gz', f'braindata/{i:02d}.nii.gz')
```
And now we can build our pipeline. The [fslpy](https://open.win.ox.ac.uk/pages/fsl/fslpy/) library has a collection of functions which we can use to call the FSL commands that we need. We need to do a little work, as by default `dask.delayed` assumes that the dependencies of a task are passed in as arguments (there are other ways of dealing with this, but in this example it is easy to simply write a few small functions that define each of our tasks):
And now we can build our pipeline. The [fslpy](https://open.win.ox.ac.uk/pages/fsl/fslpy/) library has a collection of functions which we can use to call the FSL commands that we need.
We need to do a little work, as by default `dask.delayed` assumes that the dependencies of a function (i.e. other `delayed` functions) are passed to the function as arguments. There are [other methods](https://docs.dask.org/en/stable/delayed.html#indirect-dependencies) of dealing with this, but often the easiest option is simply to write a few small functions that define each of our tasks, in a manner that satisfies this assumption:
```
......@@ -264,6 +352,7 @@ def bet(input, output):
return output
```
Again we use `dask.delayed` to build up a graph of tasks that need executing, starting from the input files, and ending with our final brain-extracted outputs. Again, we are not actually executing anything here - we're just building up a task graph that Dask will then execute for us later:
```
import glob
......@@ -281,13 +370,13 @@ for input in inputs:
tasks.append(b)
```
In the previous example we had a single output (the result of summing the squared input values) which we called `visualize()` on. We can also call `dask.visualize()` to visualise a collection of tasks:
In the previous example we had a single output (the result of summing the squared input values) upon which we called `visualize()`. Here we have a list of independent tasks (in the `tasks` list). However, we can still visualise them by passing the entire list to the `dask.visualize()` function:
```
dask.visualize(*tasks)
```
And, similarly, we can call `dask.compute` to run them all:
And, similarly, we can call `dask.compute` to run them all at once:
```
outputs = dask.compute(*tasks)
......@@ -299,7 +388,7 @@ print(outputs)
## Distributing computation with Dask
In the examples above, Dask was running its tasks in parallel on your local machine. But it is easy to instruct Dask to distribute its computations across multiple machines. For example, Dask has support for executing tasks on a SGE or SLURM cluster, which we have available in Oxford at FMRIB and the BDI.
In the examples above, Dask was running your tasks in parallel on your local machine. But it is easy to instruct Dask to distribute your computations across multiple machines. For example, Dask has support for executing tasks on a SGE or SLURM cluster, which we have available in Oxford at FMRIB and the BDI.
To use this functionality, we need an additional library called `dask-jobqueue` which, at the moment, is not installed as part of FSL.
......@@ -339,8 +428,9 @@ data = da.random.random((1000, 1000, 1000, 10))
print(data.mean().compute())
```
<a class="anchor" id="fsl-pipe"></a>
## fsl-pipe
# fsl-pipe
https://open.win.ox.ac.uk/pages/fsl/fsl-pipe/
......@@ -348,13 +438,13 @@ https://open.win.ox.ac.uk/pages/fsl/fsl-pipe/
fsl-pipe is a Python library (written by our very own Michiel Cottaar) which builds upon another library called [file-tree](https://open.win.ox.ac.uk/pages/fsl/file-tree/), and allows you to write an analysis pipeline in a _declarative_ manner. A pipeline is defined by:
- A "file tree" which defines the directory structure of the input and output files of the pipeline.
- A set of recipes (Python functions) describing how each of the pipeline outputs should be produced.
- A _file tree_ which defines the directory structure of the input and output files of the pipeline.
- A set of _recipes_ (Python functions) describing how each of the pipeline outputs should be produced.
In a similar vein to Dask task graphs, fsl-pipe will automatically determine which recipes should be executed, and in what order, to produce whichever output file(s) you request. Recipes can either run locally, be distributed using Dask, or be executed on a cluster using `fsl_sub`.
In a similar vein to [Dask task graphs](#low-level-dask-api), fsl-pipe will automatically determine which recipes should be executed, and in what order, to produce whichever output file(s) you request. You can instruct fsl-pipe to run your recipes locally, be parallelised or distributed using Dask, or be executed on a cluster using `fsl_sub`.
For example, let's again imagine that we have some T1 images upon which we wish to perform basic structural preprocessing, and which arranged in the following manner:
For example, let's again imagine that we have some T1 images upon which we wish to perform basic structural preprocessing, and which are arranged in the following manner:
> ```
> subjectA/
......@@ -365,28 +455,157 @@ For example, let's again imagine that we have some T1 images upon which we wish
> T1w.nii.gz
> ```
The code cell below will automatically create a dummy data set with the above structure:
```
import os
import shutil
for subj in 'ABC':
subjdir = f'mydata/subject{subj}'
os.makedirs(subjdir, exist_ok=True)
shutil.copy('../../applications/fslpy/bighead.nii.gz', f'{subjdir}/T1w.nii.gz')
```
We must first describe the structure of our data, and save that description as a file-tree file (e.g. `mydata.tree`). This file contains a placeholder for the subject ID, and gives both the input and any desired output files unique identifiers:
```
%%writefile mydata.tree
subject{subject}
T1w.nii.gz (t1)
T1w_brain.nii.gz (t1_brain)
T1w_fov.nii.gz (t1_fov)
T1w_reorient.nii.gz (t1_reorient)
```
Now we need to define our _recipes_, the individual processing steps that we want to perform, and that will generate our output files. This is similar to what we did above when we were using [Dask](#low-level-dask-api) - we define functions which perform each step.
```
from fsl.wrappers import bet, fslreorient2std, robustfov
from fsl_pipe import Pipeline, In, Out
def reorient(t1 : In, t1_reorient : Out):
fslreorient2std(t1, t1_reorient)
def fov(t1_reorient : In, t1_fov : Out):
robustfov(t1_reorient, t1_fov)
def brain_extract(t1_fov : In, t1_brain : Out):
bet(t1_fov, t1_brain)
```
Note that we have also annotated the inputs and outputs of each function, and have used the same identifiers that we used in our `mydata.tree` file above. By doing this, `fsl-pipe` is automatically able to determine the steps that would be required in order to generate the output files that we request.
Once we have defined all of our recipes, we need to create a `Pipeline` object, and add all of our functions to it:
```
pipe = Pipeline()
pipe(reorient)
pipe(fov)
pipe(brain_extract)
```
> Note that it is also possible (and equivalent) to create the `Pipeline` before defining our recipe functions, and to use the pipeline as a decorator, e.g.:
>
> ```
> subject{subject}
> T1w.nii.gz (t1)
> T1w_brain.nii.gz (t1_brain)
> T1w_fov.nii.gz (t1_fov)
> T1w_reorient.nii.gz (t1_reorient)
> pipe = Pipeline()
>
> @pipe
> def reorient(t1 : In, t1_reorient : Out):
> ...
> ```
>
We now need to create a `FileTree` object which is used to generate file paths for the input and output files. The `update_glob('t1')` method instructs the `FileTree` to scan the file system, and to generate file paths for all T1 images that are present.
```
#!/usr/bin/env python
from file_tree import FileTree
tree = FileTree.read('mydata.tree', './mydata/').update_glob('t1')
```
Then it is very easy to run our pipeline:
```
jobs = pipe.generate_jobs(tree)
jobs.run()
```
The default behaviour when using fsl-pipe locally is to for one task to be executed at a time. However, if you run fsl-pipe on the cluster, it will automatically submit the jobs using `fsl_sub`. You can also tell fsl-pipe to execute the pipeline using Dask, which will cause any independent jobs to be executed in parallel, e.g.:
```
jobs.run(method='dask')
```
The above example is just one way of using fsl-pipe - the library has several powerful features, including its own command-line interface, and the ability to skip jobs for output files that already exist.
<a class="anchor" id="fsl-sub"></a>
## fsl-sub
You can use the venerable `fsl_sub` to run several tasks in parallel both on your local machine, and on a HPC cluster, such as the ones available to us at FMRIB and the BDI. `fsl_sub` is typically called from the command-line, e.g.:
> ```
> for dataset in mydata/sub-*; do
> fsl_sub ./my_processing_script.py ${dataset} --jobram 16
> done
> ```
If you are working on a local machine, `fsl_sub` will block until your command has completed. You can run multiple commands in parallel by using "array tasks" - save each of the commands you want to run to a text file, e.g.:
> ```
> echo "./my_processing_script.py mydata/sub-01" > tasks.txt
> echo "./my_processing_script.py mydata/sub-02" >> tasks.txt
> echo "./my_processing_script.py mydata/sub-03" >> tasks.txt
> echo "./my_processing_script.py mydata/sub-04" >> tasks.txt
> echo "./my_processing_script.py mydata/sub-05" >> tasks.txt
> ```
And then run `fsl_sub -t tasks.txt` - each of your commands will be executed in parallel. If you are working on a cluster, `fsl_sub` will schedule all of the commands to be executed simultaneously.
You can also call `fsl_sub` from Python by using functions from the `fslpy` library:
> ```
> from glob import glob
> from fsl.wrappers import fsl_sub
>
> for dataset in glob('mydata/sub-*'):
> fsl_sub(f'./my_processing_script.py ${dataset}', jobram=16)
> ```
And the `fslpy` wrapper functions allow you to run FSL commands with `fsl_sub`, and to specify job dependencies. For example, to submit a collection of jobs, you can pass `submit=True`:
```
from fsl.data.image import removeExt
from fsl.wrappers import bet
from glob import glob
for t1 in glob('braindata/??.nii.gz'):
t1 = removeExt(t1)
bet(t1, f'{t1}_brain', submit={'jobram':16})
```
from file_tree import FileTree
from fsl_pipe import pipe, In, Out
@pipe
def brain_extract(t1 : In, bet_output : Out, bet_mask : Out):
bet(t1, bet_output, mask=True)
You can also specify dependencies by using the ID of a previously submitted job, e.g.:
```
from fsl.data.image import removeExt
from fsl.wrappers import robustfov, bet
from glob import glob
tree = FileTree.read('data.tree').update_glob('t1')
pipe.cli(tree)
for t1 in glob('braindata/??.nii.gz'):
t1 = removeExt(t1)
jid = robustfov(t1, f'{t1}_fov', submit=True)
bet(f'{t1}_fov', f'{t1}_brain', submit={'jobram':16, 'jobhold' : jid})
```
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment