EGEE-II INFSO-RI-031688 Enabling Grids for E-sciencE EGEE08 conference, Istambul Biomed community...

EGEE-II INFSO-RI-031688

Enabling Grids for E-sciencE

www.eu-egee.org

EGEE08 conference, Istambul

Biomed community meeting

V. Breton , CNRS


EGEE-II INFSO-RI-031688 EGE08 conference, Istambul

Tuesday morning session

• Introduction (VB)• Results of survey of the life sciences community (VB)• Biomedical grid summer school (L. Milanesi)• EGI (Diana Cresti)• Perspective on EGI from life sciences (VB)



Other sessions

• Tuesday afternoon: bioinformatics– Christophe Blanchet

• Thursday morning: medical imaging and drug discovery– Johan Montagnat

• Please make sure you upload your slides for these sessions on the conference programme



Life sciences cluster

Partner name Country Person-Months

ASGC Taïwan 24

CNR-ITB Italy 18

CNRS France 90

CNU Korea 84

KISTI Korea 39

UPV Spain 18

TOTAL 273 PM



Status of cluster activities

• Support for selected services– AMGA (KISTI, UPV)– Moteur (CNRS)

• Preparation of the migration to EGI in the life sciences sector– See D. Cresti talk

• Support to application porting– Bioinformatics– Medical imaging– Drug discovery

• Cluster management



Meeting with VPH NoE

• VPH = Virtual Physiological Human– Initiative supported by EC (first call in 2008, second call in 2009)– EGEE, supporting project of VPH NoE

• Meeting at UCL with P. Coveney’s group– V. Bloch, V.B., J. Salzemann, D. Sarramia (LPC Clermont-Fd)– UCL plays a leading role in VPH NoE WP3

Design of a toolkit to access grid resources

• Discussions on possible collaboration between VPH NoE and EGEE– Use of the biomed VO– Integration of a cluster on the biomed VO– Sharing of web services to access EGEE resources– Deployment of one VPH use case on EGEE

• Next meeting this Thursday with H. Benoit-Cattin, P. Coveney, B. Jones and G. Sipos



Analysis of the needs of the French life sciences community

• Goal: participate to a multidisciplinary prospective for the national grid initiative• Format: survey circulated in April and May 2008

– 12 questions– Available online at http://www.survey

monkey.com/s.aspx?sm=vuEQtHfQu_2fPs1UUyO2aWkQ_3d_3d• Very positive community feedback

– Over 400 responses– More than 60 laboratories in 24 cities

Scientific disciplines represented in the responses

20 48

205

619929

101734

13226

Agronomie Biologie cellulaire Bioinformatique Biologie évolutive

Biologie moléculaire Biologie structurale Chimioinformatique Drug design

Ecologie, biodiversité Génomique Protéomique



Survey results (I/IV)Connaissance personnelle des grilles

0,0%

10,0%

20,0%

30,0%

40,0%

50,0%

60,0%

70,0%

80,0%

Nulle Faible Satisfaisante Etendue

Tous

Biologie

Santé

Chimioinformatique

Imagerie médicale

Professionnels de santé

Personal knowledge on grids

None Limited Satisfactory Broad

Utilisation des grilles dans les laboratoires

0,0%

10,0%

20,0%

30,0%

40,0%

50,0%

60,0%

Inexistante Anecdotique Croissante Courante

Tous

Biologie

Santé

Chimioinformatique

Imagerie médicale


Use of grids in the laboratories

None Limited Growing routinely



Survey results (II/IV)

0,0%

10,0%

20,0%

30,0%

40,0%

50,0%

60,0%

Ne sais pasFaible ou nul(<1GFlop)

Peuimportante(entre 1 et10 GFlop)

Importante(entre

10GFlop et 1TFlop)

Trésimportante(> 1 TFlop)

Besoins propres sur supercalculateurs

Tous

Biologie

Santé

Chimioinformatique

Imagerie médicale

Professionnels desanté

Besoins propres sur clusters ou grilles

0,0%10,0%20,0%30,0%40,0%50,0%60,0%

Ne sais pas Faible ounul (<10

jours CPU)

Peuimportant

(entre 10 et1 an CPU)

Important(entre 1 anet 10 ans

CPU)

Trésimportant(>10 ans

CPU)

Tous

Biologie

Santé

Chimioinformatique

Imagerie médicale


Personal need of supercomputer resources

Personal need of cluster or grid resources

Unknown Small Limited Significant Large <1GFlop [1-10GF] [10G-1TF] >1TFlop

Unknown Small Limited Significant Large <10CPUdays [10-365CPUdays] [1-10CPUyears] >10CPUyears



Survey results (III/IV)Planification des besoins de calculs

01020304050

Tous

Biolog

ie

Santé

Chimioi

nform

atique

% d

es r

épo

nse

s

très stables au cours de l'année

par pic

faciles à planifier plusieurssemaines à l'avance

difficiles à planifier

Planification des besoins de stockage

0102030405060

Tous

Biolog

ie

Santé

Chimioi

nform

atique

% d

es r

épo

nse

s

très stables au cours de l'année

par pic

faciles à planifier plusieurssemaines à l'avance

difficiles à planifier

Planning of computing needs

Planning of storage needs

All Biology Health Chemo- informatics


Very stable during the year

Very unstable with peaks

Easy to plan weeks in advance

Hard to plan

Very stable during the year

Very unstable with peaks

Easy to plan weeks in advance

Hard to plan



Survey results (IV/IV)Interface d'utilisation des ressources informatiques

0,0%

20,0%

40,0%

60,0%

80,0%

100,0%

par lignes de commande via un portail web via des applications métiers

Tous Biologie Santé Chimioinformatique Imagerie médicale Professionnels de santé

Sécurité des données en entrée

010203040506070

Tous

Biolog

ie

Santé

Chimioi

nform

atique

% d

es r

épo

nse

s

pas de contrainte de sécurité

contrôle d'accès

cryptage

anonymisation (pour lesdonnées médicales)

Security on the input and output data


No constraints

Access control

Encryption

Anonymization

User interface to grid resources

Command lines Web portal dedicated interfaces



Conclusions

• The life sciences community has homogeneous needs– Except for security, all sub-communities have very comparable

answers

• The life sciences community needs to access both cluster grids and supercomputers– Comparable needs expressed for both infrastructures– on demand computing: significant fraction of the computing needs

are difficult to plan in advance

• Significant adoption of grids by the research community– To be counterweighted by the targeted audience

• Security– 90% of the applications in biology require only access control– Only 50% for health applications, the other 50% requiring medical

data anonymization



EGI: specific thoughts for the life science SSC

• Adoption of the grid infrastructures is still in its infancy– It is critical that the biomed VO is continuously operated for the

pioneers already using the grid

• The life science community is very heterogeneous– Many sub-communities with similar requirements (see survey)– About 8 ESFRI design studies are related to life sciences

BBSRC: biobanking ELIXIR: molecular biology LIFEWATCH: biodiversity …

– Need to properly interface them to EGI

Life sciences proposed as guinea pigs of the EGI (with particle physics)



Comments on science gateways

• Development of international gateways is the duty of the research communities using it. – Interest/necessity to share some tools (workflow engines) and

technologies (web services, semantic annotation).

• SSC should coordinate the development of science gateways to guarantee interoperability and integration

• SSC should be in charge of the science gateway to the biomed VO – template for the other gateways– Development started very early in the project to be able to

distribute it to the communities



Questions

• How should the biomed community get organized?– Should there be one life sciences SSC or one per ESFRI?– If any, should biomed SSC be funded by EGI, the NGIs or the

community?

Date post:	14-Jan-2016
Category:	Documents
Upload:	blake-skinner
View:	213 times
Download:	0 times

EGEE-II INFSO-RI-031688 Enabling Grids for E-sciencE EGEE08 conference, Istambul Biomed community...

Documents