PIMS - CCP4

caddiepastData Management

Jan 31, 2013 (4 years and 9 months ago)

207 views

PIMS: The Problems of

Project Management

Robert Esnouf, Scientific Sponsor for PIMS

OPPF/STRUBI, University of Oxford

strubi

.ox.ac.uk

PIMS “mission statement”…


“To produce a commercial
-
quality freely
available laboratory information
management system (LIMS) suitable for use
in structural biology research laboratories”


Many (partially) failed efforts in the past


Process is very complex (by previous LIMS
standards)


Research processes rapidly evolve (need
configuration rather than customization)


No two laboratories have the same working
practices


Potential targets / bioinformatics annotation


Target selection and construct design


Project planning and progress


Experiments and protocols (templates)


Non
-
plate: expression, purification, “traditional” work


Plate
-
based: PCR, cloning, crystallization


QA: gels, mass spectroscopy, sequencing, DLS


Samples and sample descriptions (
e.g.

sequences)


Holders and locations


Stocks, reagents and reference data


Health and safety information


Users, roles, access / sharing and security


Databases and external references


X
-
ray diffraction / structure solution

Information to be managed…

Functionality required…


An interface for entering data


Simple to use, intuitive


Minimal client software


Secure storage of
well defined data

(database)


An interface for recovering / analyzing data


An interface for project management


Administration (configuration and management roles)


Interface to external software (
e.g.

web services)


Integration of robotic platforms


parsing output files


producing run sequence files


direct robotic control


Scientific goals for PIMS…

Recording laboratory information


A lot of data recording


10,000s of experiments


1,000,000s of samples

Data interchange and interoperation


Collaboration in protein production


Share data between stages and sites


Data transfer to beam line or NMR operations

Data mining and reporting


Analysis of positive and negative results


Data deposition


Scientific publications

The story of PIMS so far…


PIMS started as a loose consortium involving labs
in the UK, France and elsewhere


PIMS BBSRC SPoRT grant (3.62 FTE)


in collaboration with

and
in support of

other SPoRT

award holders (SSPF and MPSI) with heavy

involvement of CCP4 (2 FTE), OPPF and others


PIMS effectively started 4/2005 (one post 2/2006)


Management structure re
-
investigated late 2005


Part
-
time ‘Scientific Sponsor’ (Robert E)

who works with ‘Project Manager’ (Chris M)


Version 1.0 released 15/1/2007


Version 1.1 due 17/4/2007

PIMS version 1.0: January 2007…


Improved performance


Adequate for small
-
to
-
medium scale


Barely adequate for scale of OPPF target data


10,000 targets, 4,000 constructs imported, 3 genomes


Support for plate
-
based experiments


Simplified user interface


“Generic” interface became “Expert” interface


Development guided by end
-
user feedback


First sample tracking to link experiments together


Create a pipeline of data


Workshop to introduce users to PIMS


Now focusing on SPoRT/OPPF use

PIMS management structure…

Developer

Developer

Developer

Developer

Developer

Chris M

Line Man.

Line Man.

Project Steering

Board


Strategy &

priorities

Progress

& issues

Major feature

requests

Major feature

requests

Local issues and

requirements;

daily management

Tasks, coordination

progress monitoring

Robert E

Short
-
term / long
-
term issues…


Meeting the needs of SPoRT consortia / OPPF / YSBL
etc
.


Implementations of established experimental procedures


Interfacing existing software


Each lab gets a custom interface


Developing a truly generic LIMS for end of project


Balancing competing interests


One size fits all/no one


Model is comprehensive/cumbersome


Interface is complex


Lack of early user input


Shared goals


Common way of representing data underneath


Contributed software


Extensible application




Object Domain

Complete
Data model


Current interaction with CCPN…

PIMS model

Business Logic

User Interface

PIMS API

‘Hibernate’ API

Hibernate
Persistence Layer

PostgreSQL DB

PIMS/CCPN

Autogeneration

Software

Hibernate
Mapping Files



Review of data model/data base



ObjectDomain has ceased trading

Problems of distributed projects…


Isolated developers


Need good support


Face contradictory demands


Developers not near experimentalists


Relevance of developments


Usability of developments


Focus is provided by real use


Needs “big picture” vision to get to “real use” stage


First experience of users can be brutal


Need developers to spend time together


Code camps / teleconferencing


Email is poor communication

Problems of distributed projects…


Management by a distributed PSB


Requires consent/indulgence of collaborating groups


Hard to get PSB together for meetings


Interaction between PSB and developers


Need for clear minutes/actions


Scientific sponsor could easily be full time role


Assessment by BBSRC


Review not by computer scientists (not bad!)


Original review process contained no demo (very bad!)


Visiting group assessed PIMS in November


‘Mid
-
term’ review will consist of demo at BBSRC


PIMS non
-
plate experiments…

PIMS plate
-
based experiments…

Oxford Protein Production Facility…


Example follows 96 constructs through PCR, Gateway
cloning and expression screening with two cell lines and
two protocols:









Top shows plate usage


Bottom shows the number of 96
-
lane agarose gels, 24
-
well
colony
-
plate images and 26
-
lane SDS

PAGE gels


96 constructs uses 34 96
-
well plates and 36 24
-
well plates…


…generates 480 images of colony wells,

1536 lanes on agarose gels

and 416 lanes on SDS

PAGE gels


Target annotation
(largely covered in PIMS 0.4)


Target selection
(not planned for PIMS)


Construct design
(using VectorNTI)


Obtain/store source strain genomic DNA


Describe selected genes


Describe primers, link to VectorNTI output


Describe entry clones as plasmids


Describe expression constructs


Describe high
-
throughput expression trials


Describe solubilization trials…

Working with MPSI to increase use…

Solubilization trials (Leeds)…


Solubilization trials performed in 96
-
well format


Perform 24
-
trials per target, therefore four targets per set

Det 1

Det 2

Det 3

Det 4

Det 1

Det 2

Det 3

Det 4

Target 1

Target 3

Target 2

Target 4

Detergent concentration gradients…