DeepProg / Git / [53737a] /docs/_build/doctrees/README.doctree

Models:
AlyssaS/
DeepProg
Downloads: 1
[53737a]: / docs / _build / doctrees / README.doctree
History
Download this file
372 lines (320 with data), 46.6 kB

cdocutils.nodes
document
q)q}q(U	nametypesq}q(Xexample datasets and scriptsqNXrequirementsqNXdistributed computationqNX8creating a deepprog model using an ensemble of submodelsq	NXCcreating a simple deepprog model with one autoencoder for each omicq
NX#visualisation module (experimental)qNXDcreating a distributed deepprog model using an ensemble of submodelsqNXcontact and credentialsq
NXinstallation (local)qNXusageqNXsupport for cntk / tensorflowqNXexample scriptsqNXBsurvival integration of multi-omics using deep-learning (deepprog)qNuUsubstitution_defsq}qUparse_messagesq]qUcurrent_sourceqNU
decorationqNUautofootnote_startqKUnameidsq}q(hUexample-datasets-and-scriptsqhUrequirementsqhUdistributed-computationqh	U8creating-a-deepprog-model-using-an-ensemble-of-submodelsqh
UCcreating-a-simple-deepprog-model-with-one-autoencoder-for-each-omicq hU!visualisation-module-experimentalq!hUDcreating-a-distributed-deepprog-model-using-an-ensemble-of-submodelsq"h
Ucontact-and-credentialsq#hUinstallation-localq$hUusageq%hUsupport-for-cntk-tensorflowq&hUexample-scriptsq'hU@survival-integration-of-multi-omics-using-deep-learning-deepprogq(uUchildrenq)]q*cdocutils.nodes
section
q+)q,}q-(U	rawsourceq.UUparentq/hUsourceq0X(/home/oliver/code/SimDeep/docs/README.mdq1Utagnameq2Usectionq3U
attributesq4}q5(Udupnamesq6]Uclassesq7]Ubackrefsq8]Uidsq9]q:h(aUnamesq;]q<hauUlineq=KUdocumentq>hh)]q?(cdocutils.nodes
title
q@)qA}qB(h.XBSurvival Integration of Multi-omics using Deep-Learning (DeepProg)h/h,h0h1h2UtitleqCh4}qD(h6]h7]h8]h9]h;]uh=Kh)]qEcdocutils.nodes
Text
qFXBSurvival Integration of Multi-omics using Deep-Learning (DeepProg)qGqH}qI(h.XBSurvival Integration of Multi-omics using Deep-Learning (DeepProg)qJh/hAubaubcdocutils.nodes
paragraph
qK)qL}qM(h.XQThis package allows to combine multi-omics data together with survival. Using autoencoders, the pipeline creates new features and identify those linked with survival, using CoxPH regression.
The omic data used in the original study are RNA-Seq, MiR and Methylation. However, this approach can be extended to any combination of omic data.h/h,h0h1h2U	paragraphqNh4}qO(h6]h7]h8]h9]h;]uh=Kh>hh)]qP(hFX¾This package allows to combine multi-omics data together with survival. Using autoencoders, the pipeline creates new features and identify those linked with survival, using CoxPH regression.qQqR}qS(h.X¾This package allows to combine multi-omics data together with survival. Using autoencoders, the pipeline creates new features and identify those linked with survival, using CoxPH regression.qTh0Nh=Nh>hh/hLubhFX
qU}qV(h.Uh0Nh=Nh>hh/hLubhFXThe omic data used in the original study are RNA-Seq, MiR and Methylation. However, this approach can be extended to any combination of omic data.qWqX}qY(h.XThe omic data used in the original study are RNA-Seq, MiR and Methylation. However, this approach can be extended to any combination of omic data.qZh0Nh=Nh>hh/hLubeubhK)q[}q\(h.XThe current package contains the omic data used in the study and a copy of the model computed. However, it is very easy to recreate a new model from scratch using any combination of omic data.
The omic data and the survival files should be in tsv (Tabular Separated Values) format and examples are provided. The deep-learning framework uses Keras, which is a embedding of Theano / tensorflow/ CNTK.h/h,h0h1h2hNh4}q](h6]h7]h8]h9]h;]uh=Kh>hh)]q^(hFXÀThe current package contains the omic data used in the study and a copy of the model computed. However, it is very easy to recreate a new model from scratch using any combination of omic data.q_q`}qa(h.XÀThe current package contains the omic data used in the study and a copy of the model computed. However, it is very easy to recreate a new model from scratch using any combination of omic data.qbh0Nh=Nh>hh/h[ubhFX
qc}qd(h.Uh0Nh=Nh>hh/h[ubhFXÍThe omic data and the survival files should be in tsv (Tabular Separated Values) format and examples are provided. The deep-learning framework uses Keras, which is a embedding of Theano / tensorflow/ CNTK.qeqf}qg(h.XÍThe omic data and the survival files should be in tsv (Tabular Separated Values) format and examples are provided. The deep-learning framework uses Keras, which is a embedding of Theano / tensorflow/ CNTK.qhh0Nh=Nh>hh/h[ubeubh+)qi}qj(h.Uh/h,h0h1h2h3h4}qk(h6]h7]h8]h9]qlhah;]qmhauh=K
h>hh)]qn(h@)qo}qp(h.XRequirementsh/hih0h1h2hCh4}qq(h6]h7]h8]h9]h;]uh=K
h)]qrhFXRequirementsqsqt}qu(h.XRequirementsqvh/houbaubcdocutils.nodes
bullet_list
qw)qx}qy(h.Uh/hih0h1h2Ubullet_listqzh4}q{(h6]h7]h8]h9]h;]uh=Kh>hh)]q|(cdocutils.nodes
list_item
q})q~}q(h.Uh/hxh0h1h2U	list_itemqh4}q(h6]h7]h8]h9]h;]uh=Kh>hh)]qhK)q}q(h.X
Python 2 or 3h/h~h0h1h2hNh4}q(h6]h7]h8]h9]h;]uh=Kh>hh)]qhFX
Python 2 or 3qq}q(h.X
Python 2 or 3qh0Nh=Nh>hh/hubaubaubh})q}q(h.Uh/hxh0h1h2hh4}q(h6]h7]h8]h9]h;]uh=Kh>hh)]qhK)q}q(h.X6theano (the used version for the manuscript was 0.8.2)h/hh0h1h2hNh4}q(h6]h7]h8]h9]h;]uh=Kh>hh)]q(cdocutils.nodes
reference
q)q}q(h.Xtheanoh/hh0h1h2U	referenceqh4}q(UrefuriqU4http://deeplearning.net/software/theano/install.htmlqh9]h8]h6]h7]h;]uh=Kh>hh)]qhFXtheanoqq}q(h.Xtheanoqh0Nh=Nh>hh/hubaubhFX0 (the used version for the manuscript was 0.8.2)qq }q¡(h.X0 (the used version for the manuscript was 0.8.2)q¢h0Nh=Nh>hh/hubeubaubh})q£}q¤(h.Uh/hxh0h1h2hh4}q¥(h6]h7]h8]h9]h;]uh=K
h>hh)]q¦hK)q§}q¨(h.X1tensorflow as a more robust alternative to theanoh/h£h0h1h2hNh4}q©(h6]h7]h8]h9]h;]uh=K
h>hh)]qª(h)q«}q¬(h.X
tensorflowh/h§h0h1h2hh4}q(hUhttps://www.tensorflow.org/q®h9]h8]h6]h7]h;]uh=K
h>hh)]q¯hFX
tensorflowq°q±}q²(h.X
tensorflowq³h0Nh=Nh>hh/h«ubaubhFX' as a more robust alternative to theanoq´qµ}q¶(h.X' as a more robust alternative to theanoq·h0Nh=Nh>hh/h§ubeubaubh})q¸}q¹(h.Uh/hxh0h1h2hh4}qº(h6]h7]h8]h9]h;]uh=Kh>hh)]q»hK)q¼}q½(h.Xcntk CNTK is anoter DL library that can present some advantages compared to tensorflow or theano. See https://docs.microsoft.com/en-us/cognitive-toolkit/h/h¸h0h1h2hNh4}q¾(h6]h7]h8]h9]h;]uh=Kh>hh)]q¿(h)qÀ}qÁ(h.Xcntkh/h¼h0h1h2hh4}qÂ(hU!https://github.com/microsoft/CNTKqÃh9]h8]h6]h7]h;]uh=Kh>hh)]qÄhFXcntkqÅqÆ}qÇ(h.XcntkqÈh0Nh=Nh>hh/hÀubaubhFXb CNTK is anoter DL library that can present some advantages compared to tensorflow or theano. See qÉqÊ}qË(h.Xb CNTK is anoter DL library that can present some advantages compared to tensorflow or theano. See qÌh0Nh=Nh>hh/h¼ubh)qÍ}qÎ(h.X3https://docs.microsoft.com/en-us/cognitive-toolkit/h/h¼h0h1h2hh4}qÏ(hU3https://docs.microsoft.com/en-us/cognitive-toolkit/qÐh9]h8]h6]h7]h;]uh=Kh>hh)]qÑhFX3https://docs.microsoft.com/en-us/cognitive-toolkit/qÒqÓ}qÔ(h.X3https://docs.microsoft.com/en-us/cognitive-toolkit/qÕh0Nh=Nh>hh/hÍubaubeubaubh})qÖ}q×(h.Uh/hxh0h1h2hh4}qØ(h6]h7]h8]h9]h;]uh=Kh>hh)]qÙhK)qÚ}qÛ(h.XRh/hÖh0h1h2hNh4}qÜ(h6]h7]h8]h9]h;]uh=Kh>hh)]qÝhFXRqÞ}qß(h.XRh0Nh=Nh>hh/hÚubaubaubh})qà}qá(h.Uh/hxh0h1h2hh4}qâ(h6]h7]h8]h9]h;]uh=Kh>hh)]qãhK)qä}qå(h.X#the R "survival" package installed.h/hàh0h1h2hNh4}qæ(h6]h7]h8]h9]h;]uh=Kh>hh)]qç(hFXthe R qèqé}qê(h.Xthe R qëh0Nh=Nh>hh/häubhFXâqì}qí(h.X"h0Nh=Nh>hh/häubhFXsurvivalqîqï}qð(h.Xsurvivalqñh0Nh=Nh>hh/häubhFXâqò}qó(h.X"h0Nh=Nh>hh/häubhFX package installed.qôqõ}qö(h.X package installed.q÷h0Nh=Nh>hh/häubeubaubh})qø}qù(h.Uh/hxh0h1h2hh4}qú(h6]h7]h8]h9]h;]uh=Kh>hh)]qûhK)qü}qý(h.Xnumpy, scipyh/høh0h1h2hNh4}qþ(h6]h7]h8]h9]h;]uh=Kh>hh)]qÿhFXnumpy, scipyrr}r(h.Xnumpy, scipyrh0Nh=Nh>hh/hüubaubaubh})r}r(h.Uh/hxh0h1h2hh4}r(h6]h7]h8]h9]h;]uh=Kh>hh)]rhK)r}r	(h.Xscikit-learn (>=0.18)h/jh0h1h2hNh4}r
(h6]h7]h8]h9]h;]uh=Kh>hh)]rhFXscikit-learn (>=0.18)rr
}r(h.Xscikit-learn (>=0.18)rh0Nh=Nh>hh/jubaubaubh})r}r(h.Uh/hxh0h1h2hh4}r(h6]h7]h8]h9]h;]uh=Kh>hh)]rhK)r}r(h.X¾rpy2 2.8.6 (for python2 rpy2 can be install with: pip install rpy2==2.8.6, for python3 pip3 install rpy2==2.8.6). It seems that newer version of rpy2 might not work due to a bug (not tested)h/jh0h1h2hNh4}r(h6]h7]h8]h9]h;]uh=Kh>hh)]rhFX¾rpy2 2.8.6 (for python2 rpy2 can be install with: pip install rpy2==2.8.6, for python3 pip3 install rpy2==2.8.6). It seems that newer version of rpy2 might not work due to a bug (not tested)rr}r(h.X¾rpy2 2.8.6 (for python2 rpy2 can be install with: pip install rpy2==2.8.6, for python3 pip3 install rpy2==2.8.6). It seems that newer version of rpy2 might not work due to a bug (not tested)rh0Nh=Nh>hh/jubaubaubeubcdocutils.nodes
literal_block
r)r}r(h.Xÿpip install theano --user # Original backend used OR
pip install tensorflow --user # Alternative backend for keras supposely for efficient
pip install keras --user
pip install rpy2==2.8.6 --user

#If you want to use theano or CNTK
nano ~/.keras/keras.jsonh/hih0h1h2U
literal_blockrh4}r (UlanguageXbashr!U	xml:spacer"Upreserver#h9]h8]h6]h7]h;]uh=Kh>hh)]r$hFXÿpip install theano --user # Original backend used OR
pip install tensorflow --user # Alternative backend for keras supposely for efficient
pip install keras --user
pip install rpy2==2.8.6 --user

#If you want to use theano or CNTK
nano ~/.keras/keras.jsonr%r&}r'(h.Uh/jubaubhw)r(}r)(h.Uh/hih0h1h2hzh4}r*(h6]h7]h8]h9]h;]uh=K h>hh)]r+h})r,}r-(h.Uh/j(h0h1h2hh4}r.(h6]h7]h8]h9]h;]uh=K h>hh)]r/hK)r0}r1(h.XR installationh/j,h0h1h2hNh4}r2(h6]h7]h8]h9]h;]uh=K h>hh)]r3hFXR installationr4r5}r6(h.XR installationr7h0Nh=Nh>hh/j0ubaubaubaubj)r8}r9(h.Xxinstall.package("survival")
install.package("glmnet")
source("https://bioconductor.org/biocLite.R")
biocLite("survcomp")h/hih0h1h2jh4}r:(UlanguageXRj"j#h9]h8]h6]h7]h;]uh=Kh>hh)]r;hFXxinstall.package("survival")
install.package("glmnet")
source("https://bioconductor.org/biocLite.R")
biocLite("survcomp")r<r=}r>(h.Uh/j8ubaubh+)r?}r@(h.Uh/hih0h1h2h3h4}rA(h6]h7]h8]h9]rBh&ah;]rChauh=K*h>hh)]rD(h@)rE}rF(h.XSupport for CNTK / tensorflowh/j?h0h1h2hCh4}rG(h6]h7]h8]h9]h;]uh=K*h)]rHhFXSupport for CNTK / tensorflowrIrJ}rK(h.XSupport for CNTK / tensorflowrLh/jEubaubhw)rM}rN(h.Uh/j?h0h1h2hzh4}rO(h6]h7]h8]h9]h;]uh=K+h>hh)]rPh})rQ}rR(h.Uh/jMh0h1h2hh4}rS(h6]h7]h8]h9]h;]uh=K+h>hh)]rThK)rU}rV(h.XgWe originally used Keras with theano as backend plateform. However, Tensorflow or CNTK are more recent DL framework that can be faster or more stable than theano. Because keras supports these 3 backends, it is possible to use them as alternative to theano. To change backend, please configure the $HOME/.keras/keras.json file. (See official instruction here).h/jQh0h1h2hNh4}rW(h6]h7]h8]h9]h;]uh=K+h>hh)]rX(hFXDWe originally used Keras with theano as backend plateform. However, rYrZ}r[(h.XDWe originally used Keras with theano as backend plateform. However, r\h0Nh=Nh>hh/jUubh)r]}r^(h.X
Tensorflowh/jUh0h1h2hh4}r_(hUhttps://www.tensorflow.org/r`h9]h8]h6]h7]h;]uh=K+h>hh)]rahFX
Tensorflowrbrc}rd(h.X
Tensorflowreh0Nh=Nh>hh/j]ubaubhFX or rfrg}rh(h.X or rih0Nh=Nh>hh/jUubh)rj}rk(h.XCNTKh/jUh0h1h2hh4}rl(hU3https://docs.microsoft.com/en-us/cognitive-toolkit/rmh9]h8]h6]h7]h;]uh=K+h>hh)]rnhFXCNTKrorp}rq(h.XCNTKrrh0Nh=Nh>hh/jjubaubhFXÓ are more recent DL framework that can be faster or more stable than theano. Because keras supports these 3 backends, it is possible to use them as alternative to theano. To change backend, please configure the rsrt}ru(h.XÓ are more recent DL framework that can be faster or more stable than theano. Because keras supports these 3 backends, it is possible to use them as alternative to theano. To change backend, please configure the rvh0Nh=Nh>hh/jUubcdocutils.nodes
literal
rw)rx}ry(h.X$HOME/.keras/keras.jsonrzh/jUh0h1h2Uliteralr{h4}r|(h6]h7]h8]h9]h;]uh=Kh>hh)]r}hFX$HOME/.keras/keras.jsonr~r}r(h.Uh0Nh=Nh>hh/jxubaubhFX! file. (See official instruction rr}r(h.X! file. (See official instruction rh0Nh=Nh>hh/jUubh)r}r(h.Xhereh/jUh0h1h2hh4}r(hUhttps://keras.io/backend/rh9]h8]h6]h7]h;]uh=K+h>hh)]rhFXhererr}r(h.Xhererh0Nh=Nh>hh/jubaubhFX).rr}r(h.X).rh0Nh=Nh>hh/jUubeubaubaubhK)r}r(h.X/The default configuration file looks like this:h/j?h0h1h2hNh4}r(h6]h7]h8]h9]h;]uh=K-h>hh)]rhFX/The default configuration file looks like this:rr}r(h.X/The default configuration file looks like this:rh0Nh=Nh>hh/jubaubj)r}r(h.Xx{
    "image_data_format": "channels_last",
    "epsilon": 1e-07,
    "floatx": "float32",
    "backend": "tensorflow"
}h/j?h0h1h2jh4}r(UlanguageXjsonrj"j#h9]h8]h6]h7]h;]uh=Kh>hh)]rhFXx{
    "image_data_format": "channels_last",
    "epsilon": 1e-07,
    "floatx": "float32",
    "backend": "tensorflow"
}rr }r¡(h.Uh/jubaubeubeubh+)r¢}r£(h.Uh/h,h0h1h2h3h4}r¤(h6]h7]h8]h9]r¥hah;]r¦hauh=K8h>hh)]r§(h@)r¨}r©(h.XDistributed computationh/j¢h0h1h2hCh4}rª(h6]h7]h8]h9]h;]uh=K8h)]r«hFXDistributed computationr¬r}r®(h.XDistributed computationr¯h/j¨ubaubhw)r°}r±(h.Uh/j¢h0h1h2hzh4}r²(h6]h7]h8]h9]h;]uh=K9h>hh)]r³(h})r´}rµ(h.Uh/j°h0h1h2hh4}r¶(h6]h7]h8]h9]h;]uh=K9h>hh)]r·hK)r¸}r¹(h.XÙIt is possible to use the python ray framework https://github.com/ray-project/ray to control the parallel computation of the multiple models. To use this framework, it is required to install it: pip install ray --userh/j´h0h1h2hNh4}rº(h6]h7]h8]h9]h;]uh=K9h>hh)]r»(hFX/It is possible to use the python ray framework r¼r½}r¾(h.X/It is possible to use the python ray framework r¿h0Nh=Nh>hh/j¸ubh)rÀ}rÁ(h.X"https://github.com/ray-project/rayh/j¸h0h1h2hh4}rÂ(hU"https://github.com/ray-project/rayrÃh9]h8]h6]h7]h;]uh=K9h>hh)]rÄhFX"https://github.com/ray-project/rayrÅrÆ}rÇ(h.X"https://github.com/ray-project/rayrÈh0Nh=Nh>hh/jÀubaubhFXr to control the parallel computation of the multiple models. To use this framework, it is required to install it: rÉrÊ}rË(h.Xr to control the parallel computation of the multiple models. To use this framework, it is required to install it: rÌh0Nh=Nh>hh/j¸ubjw)rÍ}rÎ(h.Xpip install ray --userrÏh/j¸h0h1h2j{h4}rÐ(h6]h7]h8]h9]h;]uh=Kh>hh)]rÑhFXpip install ray --userrÒrÓ}rÔ(h.Uh0Nh=Nh>hh/jÍubaubeubaubh})rÕ}rÖ(h.Uh/j°h0h1h2hh4}r×(h6]h7]h8]h9]h;]uh=K:h>hh)]rØhK)rÙ}rÚ(h.XgAlternatively, it is also possible to create the model one by one without the need of the ray frameworkh/jÕh0h1h2hNh4}rÛ(h6]h7]h8]h9]h;]uh=K:h>hh)]rÜhFXgAlternatively, it is also possible to create the model one by one without the need of the ray frameworkrÝrÞ}rß(h.XgAlternatively, it is also possible to create the model one by one without the need of the ray frameworkràh0Nh=Nh>hh/jÙubaubaubeubeubh+)rá}râ(h.Uh/h,h0h1h2h3h4}rã(h6]h7]h8]h9]räh!ah;]råhauh=K<h>hh)]ræ(h@)rç}rè(h.X#Visualisation module (Experimental)h/jáh0h1h2hCh4}ré(h6]h7]h8]h9]h;]uh=K<h)]rêhFX#Visualisation module (Experimental)rërì}rí(h.X#Visualisation module (Experimental)rîh/jçubaubhw)rï}rð(h.Uh/jáh0h1h2hzh4}rñ(h6]h7]h8]h9]h;]uh=K=h>hh)]rò(h})ró}rô(h.Uh/jïh0h1h2hh4}rõ(h6]h7]h8]h9]h;]uh=K=h>hh)]röhK)r÷}rø(h.XTo visualise test sets projected into the multi-omic survival space, it is required to install mpld3 module: pip install mpld3 --userh/jóh0h1h2hNh4}rù(h6]h7]h8]h9]h;]uh=K=h>hh)]rú(hFX_To visualise test sets projected into the multi-omic survival space, it is required to install rûrü}rý(h.X_To visualise test sets projected into the multi-omic survival space, it is required to install rþh0Nh=Nh>hh/j÷ubjw)rÿ}r(h.Xmpld3rh/j÷h0h1h2j{h4}r(h6]h7]h8]h9]h;]uh=Kh>hh)]rhFXmpld3rr}r(h.Uh0Nh=Nh>hh/jÿubaubhFX	 module: rr}r	(h.X	 module: r
h0Nh=Nh>hh/j÷ubjw)r}r(h.Xpip install mpld3 --userr
h/j÷h0h1h2j{h4}r(h6]h7]h8]h9]h;]uh=Kh>hh)]rhFXpip install mpld3 --userrr}r(h.Uh0Nh=Nh>hh/jubaubeubaubh})r}r(h.Uh/jïh0h1h2hh4}r(h6]h7]h8]h9]h;]uh=K>h>hh)]rhK)r}r(h.X#Note that the pip version of mpld3 installed on my computer presented a bug: TypeError: array([1.]) is not JSON serializable . However, the newest version of the mpld3 available from the github solved this issue. It is therefore recommended to install the newest version to avoid this issue.h/jh0h1h2hNh4}r(h6]h7]h8]h9]h;]uh=K>h>hh)]r(hFXHNote that the pip version of mpld3 installed on my computer presented a rr}r(h.XHNote that the pip version of mpld3 installed on my computer presented a rh0Nh=Nh>hh/jubh)r}r (h.Xbugh/jh0h1h2hh4}r!(hU)https://github.com/mpld3/mpld3/issues/434r"h9]h8]h6]h7]h;]uh=K>h>hh)]r#hFXbugr$r%}r&(h.Xbugr'h0Nh=Nh>hh/jubaubhFX: r(r)}r*(h.X: r+h0Nh=Nh>hh/jubjw)r,}r-(h.X0TypeError: array([1.]) is not JSON serializable r.h/jh0h1h2j{h4}r/(h6]h7]h8]h9]h;]uh=Kh>hh)]r0hFX0TypeError: array([1.]) is not JSON serializable r1r2}r3(h.Uh0Nh=Nh>hh/j,ubaubhFX. However, the r4r5}r6(h.X. However, the r7h0Nh=Nh>hh/jubh)r8}r9(h.Xnewesth/jh0h1h2hh4}r:(hUhttps://github.com/mpld3/mpld3r;h9]h8]h6]h7]h;]uh=K>h>hh)]r<hFXnewestr=r>}r?(h.Xnewestr@h0Nh=Nh>hh/j8ubaubhFX version of the mpld3 available from the github solved this issue. It is therefore recommended to install the newest version to avoid this issue.rArB}rC(h.X version of the mpld3 available from the github solved this issue. It is therefore recommended to install the newest version to avoid this issue.rDh0Nh=Nh>hh/jubeubaubeubeubh+)rE}rF(h.Uh/h,h0h1h2h3h4}rG(h6]h7]h8]h9]rHh$ah;]rIhauh=K@h>hh)]rJ(h@)rK}rL(h.Xinstallation (local)h/jEh0h1h2hCh4}rM(h6]h7]h8]h9]h;]uh=K@h)]rNhFXinstallation (local)rOrP}rQ(h.Xinstallation (local)rRh/jKubaubj)rS}rT(h.Xfgit clone https://github.com/lanagarmire/SimDeep.git
cd SimDeep
pip install -r requirements.txt --userh/jEh0h1h2jh4}rU(UlanguageXbashrVj"j#h9]h8]h6]h7]h;]uh=Kh>hh)]rWhFXfgit clone https://github.com/lanagarmire/SimDeep.git
cd SimDeep
pip install -r requirements.txt --userrXrY}rZ(h.Uh/jSubaubeubh+)r[}r\(h.Uh/h,h0h1h2h3h4}r](h6]h7]h8]h9]r^h%ah;]r_hauh=KHh>hh)]r`(h@)ra}rb(h.XUsageh/j[h0h1h2hCh4}rc(h6]h7]h8]h9]h;]uh=KHh)]rdhFXUsagererf}rg(h.XUsagerhh/jaubaubhw)ri}rj(h.Uh/j[h0h1h2hzh4}rk(h6]h7]h8]h9]h;]uh=KIh>hh)]rlh})rm}rn(h.Uh/jih0h1h2hh4}ro(h6]h7]h8]h9]h;]uh=KIh>hh)]rphK)rq}rr(h.XItest if simdeep is functional (all the software are correctly installed):h/jmh0h1h2hNh4}rs(h6]h7]h8]h9]h;]uh=KIh>hh)]rthFXItest if simdeep is functional (all the software are correctly installed):rurv}rw(h.XItest if simdeep is functional (all the software are correctly installed):rxh0Nh=Nh>hh/jqubaubaubaubj)ry}rz(h.Xs  python test/test_dummy_boosting_stacking.py -v # OR
  nosetests test -v # Improved version of python unit testingh/j[h0h1h2jh4}r{(UlanguageXbashr|j"j#h9]h8]h6]h7]h;]uh=Kh>hh)]r}hFXs  python test/test_dummy_boosting_stacking.py -v # OR
  nosetests test -v # Improved version of python unit testingr~r}r(h.Uh/jyubaubhw)r}r(h.Uh/j[h0h1h2hzh4}r(h6]h7]h8]h9]h;]uh=KPh>hh)]rh})r}r(h.Uh/jh0h1h2hh4}r(h6]h7]h8]h9]h;]uh=KPh>hh)]r(hK)r}r(h.XAll the default parameters are defined in the config file: ./simdeep/config.py but can be passed dynamically. Three types of parameters must be defined:h/jh0h1h2hNh4}r(h6]h7]h8]h9]h;]uh=KPh>hh)]r(hFX;All the default parameters are defined in the config file: rr}r(h.X;All the default parameters are defined in the config file: rh0Nh=Nh>hh/jubjw)r}r(h.X./simdeep/config.pyrh/jh0h1h2j{h4}r(h6]h7]h8]h9]h;]uh=Kh>hh)]rhFX./simdeep/config.pyrr}r(h.Uh0Nh=Nh>hh/jubaubhFXJ but can be passed dynamically. Three types of parameters must be defined:rr}r(h.XJ but can be passed dynamically. Three types of parameters must be defined:rh0Nh=Nh>hh/jubeubhw)r}r(h.Uh/jh0h1h2hzh4}r(h6]h7]h8]h9]h;]uh=KQh>hh)]r (h})r¡}r¢(h.Uh/jh0h1h2hh4}r£(h6]h7]h8]h9]h;]uh=KQh>hh)]r¤(hK)r¥}r¦(h.X3The training dataset (omics + survival input files)h/j¡h0h1h2hNh4}r§(h6]h7]h8]h9]h;]uh=KQh>hh)]r¨hFX3The training dataset (omics + survival input files)r©rª}r«(h.X3The training dataset (omics + survival input files)r¬h0Nh=Nh>hh/j¥ubaubhw)r}r®(h.Uh/j¡h0h1h2hzh4}r¯(h6]h7]h8]h9]h;]uh=KRh>hh)]r°h})r±}r²(h.Uh/jh0h1h2hh4}r³(h6]h7]h8]h9]h;]uh=KRh>hh)]r´hK)rµ}r¶(h.XXIn addition, the parameters of the test set, i.e. the omic dataset and the survival fileh/j±h0h1h2hNh4}r·(h6]h7]h8]h9]h;]uh=KRh>hh)]r¸hFXXIn addition, the parameters of the test set, i.e. the omic dataset and the survival filer¹rº}r»(h.XXIn addition, the parameters of the test set, i.e. the omic dataset and the survival filer¼h0Nh=Nh>hh/jµubaubaubaubeubh})r½}r¾(h.Uh/jh0h1h2hh4}r¿(h6]h7]h8]h9]h;]uh=KSh>hh)]rÀhK)rÁ}rÂ(h.X[The parameters of the autoencoder (the default parameters works but it might be fine-tuned.h/j½h0h1h2hNh4}rÃ(h6]h7]h8]h9]h;]uh=KSh>hh)]rÄhFX[The parameters of the autoencoder (the default parameters works but it might be fine-tuned.rÅrÆ}rÇ(h.X[The parameters of the autoencoder (the default parameters works but it might be fine-tuned.rÈh0Nh=Nh>hh/jÁubaubaubh})rÉ}rÊ(h.Uh/jh0h1h2hh4}rË(h6]h7]h8]h9]h;]uh=KTh>hh)]rÌhK)rÍ}rÎ(h.XHThe parameters of the classification procedures (default are still good)h/jÉh0h1h2hNh4}rÏ(h6]h7]h8]h9]h;]uh=KTh>hh)]rÐhFXHThe parameters of the classification procedures (default are still good)rÑrÒ}rÓ(h.XHThe parameters of the classification procedures (default are still good)rÔh0Nh=Nh>hh/jÍubaubaubeubeubaubeubh+)rÕ}rÖ(h.Uh/h,h0h1h2h3h4}r×(h6]h7]h8]h9]rØhah;]rÙhauh=KWh>hh)]rÚ(h@)rÛ}rÜ(h.XExample datasets and scriptsh/jÕh0h1h2hCh4}rÝ(h6]h7]h8]h9]h;]uh=KWh)]rÞhFXExample datasets and scriptsrßrà}rá(h.XExample datasets and scriptsrâh/jÛubaubhK)rã}rä(h.X(An omic .tsv file must have this format:h/jÕh0h1h2hNh4}rå(h6]h7]h8]h9]h;]uh=KXh>hh)]ræhFX(An omic .tsv file must have this format:rçrè}ré(h.X(An omic .tsv file must have this format:rêh0Nh=Nh>hh/jãubaubj)rë}rì(h.X]head mir_dummy.tsv

Samples        dummy_mir_0     dummy_mir_1     dummy_mir_2     dummy_mir_3 ...
sample_test_0  0.469656032287  0.347987447237  0.706633335508  0.440068758445 ...
sample_test_1  0.0453108219657 0.0234642968791 0.593393816691  0.981872970341 ...
sample_test_2  0.908784043793  0.854397550009  0.575879144667  0.553333958713 ...
...
h/jÕh0h1h2jh4}rí(UlanguageXbashrîj"j#h9]h8]h6]h7]h;]uh=Kh>hh)]rïhFX]head mir_dummy.tsv

Samples        dummy_mir_0     dummy_mir_1     dummy_mir_2     dummy_mir_3 ...
sample_test_0  0.469656032287  0.347987447237  0.706633335508  0.440068758445 ...
sample_test_1  0.0453108219657 0.0234642968791 0.593393816691  0.981872970341 ...
sample_test_2  0.908784043793  0.854397550009  0.575879144667  0.553333958713 ...
...
rðrñ}rò(h.Uh/jëubaubhK)ró}rô(h.X&a survival file must have this format:h/jÕh0h1h2hNh4}rõ(h6]h7]h8]h9]h;]uh=Keh>hh)]röhFX&a survival file must have this format:r÷rø}rù(h.X&a survival file must have this format:rúh0Nh=Nh>hh/jóubaubj)rû}rü(h.Xhead survival_dummy.tsv

Samples        days event
sample_test_0  134  1
sample_test_1  291  0
sample_test_2  125  1
sample_test_3  43   0
...
h/jÕh0h1h2jh4}rý(UlanguageXbashrþj"j#h9]h8]h6]h7]h;]uh=Kh>hh)]rÿhFXhead survival_dummy.tsv

Samples        days event
sample_test_0  134  1
sample_test_1  291  0
sample_test_2  125  1
sample_test_3  43   0
...
rr}r(h.Uh/jûubaubhK)r}r(h.X&As examples, we included two datasets:h/jÕh0h1h2hNh4}r(h6]h7]h8]h9]h;]uh=Ksh>hh)]rhFX&As examples, we included two datasets:rr}r	(h.X&As examples, we included two datasets:r
h0Nh=Nh>hh/jubaubhw)r}r(h.Uh/jÕh0h1h2hzh4}r
(h6]h7]h8]h9]h;]uh=Kth>hh)]rh})r}r(h.Uh/jh0h1h2hh4}r(h6]h7]h8]h9]h;]uh=Kth>hh)]rhK)r}r(h.X4A dummy example dataset in the example/data/ folder:h/jh0h1h2hNh4}r(h6]h7]h8]h9]h;]uh=Kth>hh)]r(hFXA dummy example dataset in the rr}r(h.XA dummy example dataset in the rh0Nh=Nh>hh/jubjw)r}r(h.X
example/data/rh/jh0h1h2j{h4}r(h6]h7]h8]h9]h;]uh=Kh>hh)]rhFX
example/data/r r!}r"(h.Uh0Nh=Nh>hh/jubaubhFX folder:r#r$}r%(h.X folder:r&h0Nh=Nh>hh/jubeubaubaubj)r'}r((h.Xìexamples
âââ data
âÂ Â  âââ meth_dummy.tsv
âÂ Â  âââ mir_dummy.tsv
âÂ Â  âââ rna_dummy.tsv
âÂ Â  âââ rna_test_dummy.tsv
âÂ Â  âââ survival_dummy.tsv
âÂ Â  âââ survival_test_dummy.tsvh/jÕh0h1h2jh4}r)(UlanguageXbashr*j"j#h9]h8]h6]h7]h;]uh=Kh>hh)]r+hFXìexamples
âââ data
âÂ Â  âââ meth_dummy.tsv
âÂ Â  âââ mir_dummy.tsv
âÂ Â  âââ rna_dummy.tsv
âÂ Â  âââ rna_test_dummy.tsv
âÂ Â  âââ survival_dummy.tsv
âÂ Â  âââ survival_test_dummy.tsvr,r-}r.(h.Uh/j'ubaubhw)r/}r0(h.Uh/jÕh0h1h2hzh4}r1(h6]h7]h8]h9]h;]uh=Kh>hh)]r2h})r3}r4(h.Uh/j/h0h1h2hh4}r5(h6]h7]h8]h9]h;]uh=Kh>hh)]r6hK)r7}r8(h.XAnd a real dataset in the data folder. This dataset derives from the TCGA HCC cancer dataset. This dataset needs to be decompressed before processing:h/j3h0h1h2hNh4}r9(h6]h7]h8]h9]h;]uh=Kh>hh)]r:(hFXAnd a real dataset in the r;r<}r=(h.XAnd a real dataset in the r>h0Nh=Nh>hh/j7ubjw)r?}r@(h.XdatarAh/j7h0h1h2j{h4}rB(h6]h7]h8]h9]h;]uh=Kh>hh)]rChFXdatarDrE}rF(h.Uh0Nh=Nh>hh/j?ubaubhFXx folder. This dataset derives from the TCGA HCC cancer dataset. This dataset needs to be decompressed before processing:rGrH}rI(h.Xx folder. This dataset derives from the TCGA HCC cancer dataset. This dataset needs to be decompressed before processing:rJh0Nh=Nh>hh/j7ubeubaubaubj)rK}rL(h.X\data
âââ meth.tsv.gz
âââ mir.tsv.gz
âââ rna.tsv.gz
âââ survival.tsv
h/jÕh0h1h2jh4}rM(UlanguageXbashrNj"j#h9]h8]h6]h7]h;]uh=Kh>hh)]rOhFX\data
âââ meth.tsv.gz
âââ mir.tsv.gz
âââ rna.tsv.gz
âââ survival.tsv
rPrQ}rR(h.Uh/jKubaubeubh+)rS}rT(h.Uh/h,h0h1h2h3h4}rU(h6]h7]h8]h9]rVh ah;]rWh
auh=Kh>hh)]rX(h@)rY}rZ(h.XCCreating a simple DeepProg model with one autoencoder for each omich/jSh0h1h2hCh4}r[(h6]h7]h8]h9]h;]uh=Kh)]r\hFXCCreating a simple DeepProg model with one autoencoder for each omicr]r^}r_(h.XCCreating a simple DeepProg model with one autoencoder for each omicr`h/jYubaubhK)ra}rb(h.XôFirst, we will build a model using the example dataset from ./examples/data/ (These example files are set as default in the config.py file). We will use them to show how to construct a single DeepProg model inferring a autoencoder for each omich/jSh0h1h2hNh4}rc(h6]h7]h8]h9]h;]uh=Kh>hh)]rd(hFX<First, we will build a model using the example dataset from rerf}rg(h.X<First, we will build a model using the example dataset from rhh0Nh=Nh>hh/jaubjw)ri}rj(h.X./examples/data/rkh/jah0h1h2j{h4}rl(h6]h7]h8]h9]h;]uh=Kh>hh)]rmhFX./examples/data/rnro}rp(h.Uh0Nh=Nh>hh/jiubaubhFX¨ (These example files are set as default in the config.py file). We will use them to show how to construct a single DeepProg model inferring a autoencoder for each omicrqrr}rs(h.X¨ (These example files are set as default in the config.py file). We will use them to show how to construct a single DeepProg model inferring a autoencoder for each omicrth0Nh=Nh>hh/jaubeubj)ru}rv(h.X9
# SimDeep class can be used to build one model with one autoencoder for each omic
from simdeep.simdeep_analysis import SimDeep
from simdeep.extract_data import LoadData

help(SimDeep) # to see all the functions
help(LoadData) # to see all the functions related to loading datasets

# Defining training datasets
from simdeep.config import TRAINING_TSV
from simdeep.config import SURVIVAL_TSV

dataset = LoadData(training_tsv=TRAINING_TSV, survival_tsv=SURVIVAL_TSV)

simDeep = SimDeep(dataset=dataset) # instantiate the model with the dummy example training dataset defined in the config file
simDeep.load_training_dataset() # load the training dataset
simDeep.fit() # fit the model

# Defining test datasets
from simdeep.config import TEST_TSV
from simdeep.config import SURVIVAL_TSV_TEST

simDeep.load_new_test_dataset(TEST_TSV, SURVIVAL_TSV_TEST, fname_key='dummy')

# The test set is a dummy rna expression (generated randomly)
print(simDeep.dataset.test_tsv) # Defined in the config file
# The data type of the test set is also defined to match an existing type
print(simDeep.dataset.data_type) # Defined in the config file
simDeep.predict_labels_on_test_dataset() # Perform the classification analysis and label the set dataset

print(simDeep.test_labels)
print(simDeep.test_labels_proba)

simDeep.save_encoder('dummy_encoder.h5')
h/jSh0h1h2jh4}rw(UlanguageXpythonrxj"j#h9]h8]h6]h7]h;]uh=Kh>hh)]ryhFX9
# SimDeep class can be used to build one model with one autoencoder for each omic
from simdeep.simdeep_analysis import SimDeep
from simdeep.extract_data import LoadData

help(SimDeep) # to see all the functions
help(LoadData) # to see all the functions related to loading datasets

# Defining training datasets
from simdeep.config import TRAINING_TSV
from simdeep.config import SURVIVAL_TSV

dataset = LoadData(training_tsv=TRAINING_TSV, survival_tsv=SURVIVAL_TSV)

simDeep = SimDeep(dataset=dataset) # instantiate the model with the dummy example training dataset defined in the config file
simDeep.load_training_dataset() # load the training dataset
simDeep.fit() # fit the model

# Defining test datasets
from simdeep.config import TEST_TSV
from simdeep.config import SURVIVAL_TSV_TEST

simDeep.load_new_test_dataset(TEST_TSV, SURVIVAL_TSV_TEST, fname_key='dummy')

# The test set is a dummy rna expression (generated randomly)
print(simDeep.dataset.test_tsv) # Defined in the config file
# The data type of the test set is also defined to match an existing type
print(simDeep.dataset.data_type) # Defined in the config file
simDeep.predict_labels_on_test_dataset() # Perform the classification analysis and label the set dataset

print(simDeep.test_labels)
print(simDeep.test_labels_proba)

simDeep.save_encoder('dummy_encoder.h5')
rzr{}r|(h.Uh/juubaubeubh+)r}}r~(h.Uh/h,h0h1h2h3h4}r(h6]h7]h8]h9]rhah;]rh	auh=K³h>hh)]r(h@)r}r(h.X8Creating a DeepProg model using an ensemble of submodelsh/j}h0h1h2hCh4}r(h6]h7]h8]h9]h;]uh=K³h)]rhFX8Creating a DeepProg model using an ensemble of submodelsrr}r(h.X8Creating a DeepProg model using an ensemble of submodelsrh/jubaubhK)r}r(h.XÅSecondly, we will build a more complex DeepProg model constituted of an ensemble of sub-models each originated from a subset of the data. For that purpose, we need to use the SimDeepBoosting class:h/j}h0h1h2hNh4}r(h6]h7]h8]h9]h;]uh=Kµh>hh)]r(hFX¯Secondly, we will build a more complex DeepProg model constituted of an ensemble of sub-models each originated from a subset of the data. For that purpose, we need to use the rr}r(h.X¯Secondly, we will build a more complex DeepProg model constituted of an ensemble of sub-models each originated from a subset of the data. For that purpose, we need to use the rh0Nh=Nh>hh/jubjw)r}r(h.XSimDeepBoostingrh/jh0h1h2j{h4}r(h6]h7]h8]h9]h;]uh=Kh>hh)]rhFXSimDeepBoostingrr}r(h.Uh0Nh=Nh>hh/jubaubhFX class:rr}r(h.X class:rh0Nh=Nh>hh/jubeubj)r}r (h.X´from simdeep.simdeep_boosting import SimDeepBoosting

help(SimDeepBoosting)

from collections import OrderedDict


path_data = "../examples/data/"
# Example tsv files
tsv_files = OrderedDict([
          ('MIR', 'mir_dummy.tsv'),
          ('METH', 'meth_dummy.tsv'),
          ('RNA', 'rna_dummy.tsv'),
])

# File with survival event
survival_tsv = 'survival_dummy.tsv'

project_name = 'stacked_TestProject'
epochs = 10 # Autoencoder epochs. Other hyperparameters can be fine-tuned. See the example files
seed = 3 # random seed used for reproducibility
nb_it = 5 # This is the number of models to be fitted using only a subset of the training data
nb_threads = 2 # These treads define the number of threads to be used to compute survival function

boosting = SimDeepBoosting(
    nb_threads=nb_threads,
    nb_it=nb_it,
    split_n_fold=3,
    survival_tsv=tsv_files,
    training_tsv=survival_tsv,
    path_data=path_data,
    project_name=project_name,
    path_results=path_data,
    epochs=epochs,
    seed=seed)

# Fit the model
boosting.fit()
# Predict and write the labels
boosting.predict_labels_on_full_dataset()
# Compute internal metrics
boosting.compute_clusters_consistency_for_full_labels()
# COmpute the feature importance
boosting.compute_feature_scores_per_cluster()
# Write the feature importance
boosting.write_feature_score_per_cluster()

boosting.load_new_test_dataset(
    {'RNA': 'rna_dummy.tsv'}, # OMIC file of the test set. It doesnt have to be the same as for training
    'survival_dummy.tsv', # Survival file of the test set
    'TEST_DATA_1', # Name of the test test to be used
)

# Predict the labels on the test dataset
boosting.predict_labels_on_test_dataset()
# Compute C-index
boosting.compute_c_indexes_for_test_dataset()
# See cluster consistency
boosting.compute_clusters_consistency_for_test_labels()

# [EXPERIMENTAL] method to plot the test dataset amongst the class kernel densities
boosting.plot_supervised_kernel_for_test_sets()h/j}h0h1h2jh4}r¡(UlanguageXpythonr¢j"j#h9]h8]h6]h7]h;]uh=Kh>hh)]r£hFX´from simdeep.simdeep_boosting import SimDeepBoosting

help(SimDeepBoosting)

from collections import OrderedDict


path_data = "../examples/data/"
# Example tsv files
tsv_files = OrderedDict([
          ('MIR', 'mir_dummy.tsv'),
          ('METH', 'meth_dummy.tsv'),
          ('RNA', 'rna_dummy.tsv'),
])

# File with survival event
survival_tsv = 'survival_dummy.tsv'

project_name = 'stacked_TestProject'
epochs = 10 # Autoencoder epochs. Other hyperparameters can be fine-tuned. See the example files
seed = 3 # random seed used for reproducibility
nb_it = 5 # This is the number of models to be fitted using only a subset of the training data
nb_threads = 2 # These treads define the number of threads to be used to compute survival function

boosting = SimDeepBoosting(
    nb_threads=nb_threads,
    nb_it=nb_it,
    split_n_fold=3,
    survival_tsv=tsv_files,
    training_tsv=survival_tsv,
    path_data=path_data,
    project_name=project_name,
    path_results=path_data,
    epochs=epochs,
    seed=seed)

# Fit the model
boosting.fit()
# Predict and write the labels
boosting.predict_labels_on_full_dataset()
# Compute internal metrics
boosting.compute_clusters_consistency_for_full_labels()
# COmpute the feature importance
boosting.compute_feature_scores_per_cluster()
# Write the feature importance
boosting.write_feature_score_per_cluster()

boosting.load_new_test_dataset(
    {'RNA': 'rna_dummy.tsv'}, # OMIC file of the test set. It doesnt have to be the same as for training
    'survival_dummy.tsv', # Survival file of the test set
    'TEST_DATA_1', # Name of the test test to be used
)

# Predict the labels on the test dataset
boosting.predict_labels_on_test_dataset()
# Compute C-index
boosting.compute_c_indexes_for_test_dataset()
# See cluster consistency
boosting.compute_clusters_consistency_for_test_labels()

# [EXPERIMENTAL] method to plot the test dataset amongst the class kernel densities
boosting.plot_supervised_kernel_for_test_sets()r¤r¥}r¦(h.Uh/jubaubeubh+)r§}r¨(h.Uh/h,h0h1h2h3h4}r©(h6]h7]h8]h9]rªh"ah;]r«hauh=Køh>hh)]r¬(h@)r}r®(h.XDCreating a distributed DeepProg model using an ensemble of submodelsh/j§h0h1h2hCh4}r¯(h6]h7]h8]h9]h;]uh=Køh)]r°hFXDCreating a distributed DeepProg model using an ensemble of submodelsr±r²}r³(h.XDCreating a distributed DeepProg model using an ensemble of submodelsr´h/jubaubhK)rµ}r¶(h.X§We can allow DeepProg to distribute the creation of each submodel on different clusters/nodes/CPUs by using the ray framework.
The configuration of the nodes / clusters, or local CPUs to be used needs to be done when instanciating a new ray object with the ray API. It is however quite straightforward to define the number of instances launched on a local machine such as in the example below in which 3 instances are used.h/j§h0h1h2hNh4}r·(h6]h7]h8]h9]h;]uh=Kúh>hh)]r¸(hFX~We can allow DeepProg to distribute the creation of each submodel on different clusters/nodes/CPUs by using the ray framework.r¹rº}r»(h.X~We can allow DeepProg to distribute the creation of each submodel on different clusters/nodes/CPUs by using the ray framework.r¼h0Nh=Nh>hh/jµubhFX
r½}r¾(h.Uh0Nh=Nh>hh/jµubhFXThe configuration of the nodes / clusters, or local CPUs to be used needs to be done when instanciating a new ray object with the ray r¿rÀ}rÁ(h.XThe configuration of the nodes / clusters, or local CPUs to be used needs to be done when instanciating a new ray object with the ray rÂh0Nh=Nh>hh/jµubh)rÃ}rÄ(h.XAPIh/jµh0h1h2hh4}rÅ(hU%https://ray.readthedocs.io/en/latest/rÆh9]h8]h6]h7]h;]uh=Kúh>hh)]rÇhFXAPIrÈrÉ}rÊ(h.XAPIrËh0Nh=Nh>hh/jÃubaubhFX. It is however quite straightforward to define the number of instances launched on a local machine such as in the example below in which 3 instances are used.rÌrÍ}rÎ(h.X. It is however quite straightforward to define the number of instances launched on a local machine such as in the example below in which 3 instances are used.rÏh0Nh=Nh>hh/jµubeubj)rÐ}rÑ(h.Xê# Instanciate a ray object that will create multiple workers
import ray
ray.init(num_cpus=3)
# More options can be used (e.g. remote clusters, AWS, memory,...etc...)
# ray can be used locally to maximize the use of CPUs on the local machine
# See ray API: https://ray.readthedocs.io/en/latest/index.html

boosting = SimDeepBoosting(
    ...
    distribute=True, # Additional option to use ray cluster scheduler
    ...
)
...
# Processing
...

# Close clusters and free memory
ray.shutdown()h/j§h0h1h2jh4}rÒ(UlanguageXpythonrÓj"j#h9]h8]h6]h7]h;]uh=Kh>hh)]rÔhFXê# Instanciate a ray object that will create multiple workers
import ray
ray.init(num_cpus=3)
# More options can be used (e.g. remote clusters, AWS, memory,...etc...)
# ray can be used locally to maximize the use of CPUs on the local machine
# See ray API: https://ray.readthedocs.io/en/latest/index.html

boosting = SimDeepBoosting(
    ...
    distribute=True, # Additional option to use ray cluster scheduler
    ...
)
...
# Processing
...

# Close clusters and free memory
ray.shutdown()rÕrÖ}r×(h.Uh/jÐubaubeubh+)rØ}rÙ(h.Uh/h,h0h1h2h3h4}rÚ(h6]h7]h8]h9]rÛh'ah;]rÜhauh=Mh>hh)]rÝ(h@)rÞ}rß(h.XExample scriptsh/jØh0h1h2hCh4}rà(h6]h7]h8]h9]h;]uh=Mh)]ráhFXExample scriptsrârã}rä(h.XExample scriptsråh/jÞubaubhK)ræ}rç(h.XzExample scripts are availables in ./examples/ which will assist you to build a model from scratch with test and real data:h/jØh0h1h2hNh4}rè(h6]h7]h8]h9]h;]uh=Mh>hh)]réhFXzExample scripts are availables in ./examples/ which will assist you to build a model from scratch with test and real data:rêrë}rì(h.XzExample scripts are availables in ./examples/ which will assist you to build a model from scratch with test and real data:ríh0Nh=Nh>hh/jæubaubj)rî}rï(h.Xaexamples
âââ create_autoencoder_from_scratch.py # Construct a simple deeprog model on the dummy example dataset
âââ example_with_dummy_data_distributed.py # Process the dummy example dataset using ray
âââ example_with_dummy_data.py # Process the dummy example dataset
âââ load_3_omics_model.py # Process the example HCC dataset

h/jØh0h1h2jh4}rð(UlanguageXbashrñj"j#h9]h8]h6]h7]h;]uh=Kh>hh)]ròhFXaexamples
âââ create_autoencoder_from_scratch.py # Construct a simple deeprog model on the dummy example dataset
âââ example_with_dummy_data_distributed.py # Process the dummy example dataset using ray
âââ example_with_dummy_data.py # Process the dummy example dataset
âââ load_3_omics_model.py # Process the example HCC dataset

rórô}rõ(h.Uh/jîubaubeubh+)rö}r÷(h.Uh/h,h0h1h2h3h4}rø(h6]h7]h8]h9]rùh#ah;]rúh
auh=M!h>hh)]rû(h@)rü}rý(h.Xcontact and credentialsh/jöh0h1h2hCh4}rþ(h6]h7]h8]h9]h;]uh=M!h)]rÿhFXcontact and credentialsrr}r(h.Xcontact and credentialsrh/jüubaubhw)r}r(h.Uh/jöh0h1h2hzh4}r(h6]h7]h8]h9]h;]uh=M"h>hh)]r(h})r}r	(h.Uh/jh0h1h2hh4}r
(h6]h7]h8]h9]h;]uh=M"h>hh)]rhK)r}r
(h.X Developer: Olivier Poirion (PhD)h/jh0h1h2hNh4}r(h6]h7]h8]h9]h;]uh=M"h>hh)]rhFX Developer: Olivier Poirion (PhD)rr}r(h.X Developer: Olivier Poirion (PhD)rh0Nh=Nh>hh/jubaubaubh})r}r(h.Uh/jh0h1h2hh4}r(h6]h7]h8]h9]h;]uh=M#h>hh)]rhK)r}r(h.X1contact: opoirion@hawaii.edu, o.poirion@gmail.comh/jh0h1h2hNh4}r(h6]h7]h8]h9]h;]uh=M#h>hh)]rhFX1contact: opoirion@hawaii.edu, o.poirion@gmail.comrr}r(h.X1contact: opoirion@hawaii.edu, o.poirion@gmail.comrh0Nh=Nh>hh/jubaubaubeubeubeubah.UUtransformerr NU
footnote_refsr!}r"Urefnamesr#}r$Usymbol_footnotesr%]r&Uautofootnote_refsr']r(Usymbol_footnote_refsr)]r*U	citationsr+]r,h>hUcurrent_liner-NUtransform_messagesr.]r/Ureporterr0NUid_startr1KU
autofootnotesr2]r3U
citation_refsr4}r5Uindirect_targetsr6]r7Usettingsr8(cdocutils.frontend
Values
r9or:}r;(Ufootnote_backlinksr<KUrecord_dependenciesr=NU
language_coder>Uenr?U	tracebackr@Upep_referencesrANUstrip_commentsrBNU
toc_backlinksrCUentryrDUrfc_base_urlrEUhttps://tools.ietf.org/html/rFU	datestamprGNUreport_levelrHKUsmartquotes_localesrI]rJU_destinationrKNU
halt_levelrLKU
strip_classesrMNhCNUerror_encoding_error_handlerrNUbackslashreplacerOUdebugrPNUembed_stylesheetrQUoutput_encoding_error_handlerrRUstrictrSU
sectnum_xformrTKUdump_transformsrUNU
docinfo_xformrVKUwarning_streamrWNUpep_file_url_templaterXUpep-%04drYUexit_status_levelrZKUconfigr[NUstrict_visitorr\NUcloak_email_addressesr]Utrim_footnote_reference_spacer^Uenvr_NUdump_pseudo_xmlr`NUexpose_internalsraNUsectsubtitle_xformrbUsource_linkrcNUrfc_referencesrdNUoutput_encodingreUutf-8rfU
source_urlrgNUinput_encodingrhU	utf-8-sigriU_disable_configrjNU	id_prefixrkUUcharacter_level_inline_markuprlU	tab_widthrmKUerror_encodingrnUUTF-8roU_sourcerph1U	generatorrqNUdump_internalsrrNUsmart_quotesrsUpep_base_urlrtU https://www.python.org/dev/peps/ruUsyntax_highlightrvUlongrwUinput_encoding_error_handlerrxjSUauto_id_prefixryUidrzUdoctitle_xformr{Ustrip_elements_with_classesr|NU
_config_filesr}]Ufile_insertion_enabledr~Uraw_enabledrKU
dump_settingsrNubUsymbol_footnote_startrKh9}r(hj¢hhih(h,h"j§h jShj}h%j[h&j?h#jöh$jEh'jØhjÕh!jáuUsubstitution_namesr}rh2h>h4}r(h6]h9]h8]Usourceh1h7]h;]uU	footnotesr]rUrefidsr}rub.