Soton Open Science Workshop 1/9/08

use mobile phone to write to the lab notebook - Jean-Claude Bradley

he's showing graphs of his lab notebook objects interconnected - Jean-Claude Bradley

there are no formal semantics in his system but can use tagging very well - Jean-Claude Bradley

I"m switching to posting comments to my feed - nobody else here :( - Jean-Claude Bradley

Well I'm here... - Richard P Grant

true - 4 people :) - Jean-Claude Bradley

Five! - Victor / Mendeley Team

six! (JCB - Can you pls link to your 'comments feed'? - mike seyfang

We recently published a paper about the transcriptome of a particular cell type. We only scratched the tip of the iceberg as to interesting analyses. Datasets are not enough, though, for publication in my mind. But you can't require all potential analyses, either, for a paper - or it's infinite. Scientific publications by necessity are about communication, but being forced to communicate all unfruitful lines of inquiry will really muddy the waters. Being *able* to is a good to have option, though. - Heather

The "paper" is good for focused communication, linking to data (which can be in other places because of format, eg protein data bank or open notebook). Science is not only about datasharing but also about communicating the ideas arising from the data (to various levels of readers). - Maxine

The point was made repeatedly yesterday that a paper has to have a 'claim'—it can't just be data. - Richard P Grant

I need to refine that definition though - I like the idea that the paper makes a 'claim' or a 'statement' describing a model - it has a definite place in (web)space and time. Then we can talk about having access to the material that supported that claim at the time it was made. - Cameron Neylon

the main problem with a paper is it generally needs to tell a complete "story" - and many perfectly good results get left out as being irrelevant to the story - Jean-Claude Bradley

On the other hand, many papers describing perfectly good data do not get into journals because the "story" told oversells what the data show. - Maxine

just putting together some slides on this theme for tomorrow - I think it cuts both ways - the people who can't get a publication and the big datasets which get big papers - lets just publish data as data and be done with it - Cameron Neylon

And I think the point comes down to this - the easiest way to make the paper well supported by the full set of data and analysis is to have a shareable lab notebook, even if it isn't shared until publication - but once you decide not to publish something then the rest can also be easily made available - Cameron Neylon

Makes sense to me, Cameron. - Maxine

Maxine - too bad we missed you yesterday...but we did meet a lot of people at Nature - Jean-Claude Bradley

I'm sorry too, I had unavoidable domestic commitments. Glad you met lots of people other than me! - Maxine

We had a fun time and met some more people, but you did miss me saying something nice about peer review :-) - Cameron Neylon

OK, Cameron, I'll have to work on getting you to repeat it next time we meet! - Maxine

"lets just publish data as data and be done with it" you write, Cameron. I agree with the "lets just publish data as data" bit - but not the "be done with it" because of quality control and rewards (jobs, etc) for two things. Communication (as in discussion above) a third - someone who understands the data needs to do this, either the author of it or someone who is pretty close to the techniques, etc, used. - Maxine

fair point. In no sense was that meant as just dump data any old how - just that what we tend to do at the moment with eg genome sequence papers server neither the journal nor the data well - Cameron Neylon from fftogo

Hypothesis + Data = Knowledge - Cameron Neylon

objects can be data or knowledge or the connections between the two? - Cameron Neylon

Ontologies connect objects - Cameron Neylon

Material = [Raw]Materiald + Method + Instrument - a general data model? - Cameron Neylon

quoting JC - -'communicate first, standardise second' - Cameron Neylon

need for clear experimental description ontologies - you still out there Frank? - Cameron Neylon

An 'Open Science' ontology? - Cameron Neylon

discussion of whether ontologies can be used in practice across domains or whether they are always limited and what that means - Cameron Neylon

wanted: onlotogies for organic chemistry reactions - Jean-Claude Bradley

Frank - yes do you have a link? - Jean-Claude Bradley

congregate around the principle of access to the full facts that support a published claim: currently there are huge problems but we all agree this is something we should try for - Richard P Grant

everybody looks drained :) - Jean-Claude Bradley

J-CB: things that don't work go into the ON too, which is (as) valuable as the things that finally make the paper... *process* - Richard P Grant

great notes Richard - Jean-Claude Bradley

'ONS Grand Challenge'—solubility in different solvents - Richard P Grant

Nature is willing to do profiles of the groups involved. Jenny wants physics & biology to also be involved... - Richard P Grant

Cameron thinks the experiment itself doesn't matter, at this marketing level. And reckons this project could then feed into biology anyway (drug discovery, after all, is biology...) - Richard P Grant

J-C wants to enforce rigour of open notebook - Richard P Grant

... which constrains the subjects that can be 'judged' by the expertise in this room, essentially. - Richard P Grant

No one has ever looked at (comprehensively) solubility in organic solvents. So it's important, too. - Richard P Grant

Bye Jim. - Richard P Grant

Does anyone know some congressmen? - Richard P Grant

well enforce that things called ONS really are - Jean-Claude Bradley

I get that J-C, but you're making the points about fields outside of yours. There has to be some kind of validation from an 'expert', yeah? - Richard P Grant

yes someone competent in a field can validate a notebook in their field - Jean-Claude Bradley

Two issues: recording data electronically, and making it open. Not the same thing - Richard P Grant

Issues of trust... trusting the data, trusting the person who put the data there... - Richard P Grant

building tools for the general public and not from the expert - Cameron Neylon

Data curation and preservation choices - disciplinary data centres, institutiona/departmental repositories, federation, national library, 'public' repositories, web archiving, commercial data store, ecosystem of hosted services, none of these, all of these - Cameron Neylon

are blogs being archived? Kind of... - Cameron Neylon

if serious stuff goes into it - what is the longevity in the longer term - Cameron Neylon

are these 'records of science'? If so how important are they. What long term preservation appraoches are in place? Funder requirements? Institutional requirement? personal requirement? - Cameron Neylon

Answer from David leahy - how far can open go? As far as possible! - Cameron Neylon

role of libraries - Richard P Grant

Alf Eaton - anything in a feed is preserved by google - Cameron Neylon

Need a good support system for the background infrastructure particularly the computational infrastructure - Cameron Neylon

An on demand workbecnh for scientists - Cameron Neylon

focus on the doing of science from the instrument to the analysis - Cameron Neylon

focus on drug discoveyr but should be applicable across a wide range of fields in the future - Cameron Neylon

managing multiple data types - viewers - services management - Cameron Neylon

workflow managment, multiple enactment engines, desktop wiorkflow editor, on demand computing - Cameron Neylon

looking for users to drive the development of specific data types etc - Cameron Neylon

he's still putting the infrastructure together - Jean-Claude Bradley

confidentiality is an option - Jean-Claude Bradley

digital signing. - Richard P Grant

not necessarily for long term storage - Jean-Claude Bradley

fine grain security control, digital signing, ambient similarity (?) - finding out what people are doing when it similar - Cameron Neylon

issue of whether you want to edit workflows on the desktop and/or use them on the web - Cameron Neylon

interested in hosting specialised services - Cameron Neylon

using an auto-QSAR process as an example - Cameron Neylon

QSAR automation - would be helpful for us - not sure when will be ready to try - Jean-Claude Bradley

a competitive work flow process where the system looks at diferent possible workflows to solve the same problem - Cameron Neylon

SHAPS looks for shape similarity - looks like LASSO - Jean-Claude Bradley

free for open science - Jean-Claude Bradley

he has funding for a few years - Jean-Claude Bradley

subscription for confidentiality and pay for use services - Cameron Neylon

presumably an option is to exploit their proximity to open data as well for value added services - Cameron Neylon

will start within next few weeks - Jean-Claude Bradley

his docking service will be freely available in a few months - that would be very useful - Jean-Claude Bradley

Are there slides of this talk available? - Rajarshi Guha

rajarshi - David is going to put them up on Slideshare I think - Cameron Neylon

concern of muddying the water (premature picture of reality) - Cameron Neylon

inspiring wild goose-chases (different between a theory in the process of directly disproved and one perpetuated in the fac eof an underlying error) e.g. small t-construct - Cameron Neylon

came from someone very famous and 'very reliable' which is itneresting - issue of trust again - Cameron Neylon

data fatigue: spend all the time looking not doing - Cameron Neylon

territorial pissing to discourage competition (big lab puts stuff out to keep smaller groups out) - Cameron Neylon

and they might piss just to scare people away even though they actually have no intention of working on it - Richard P Grant

slave to fashion - swept along by a prevailing view - would this be amplified? - Cameron Neylon

redundancy - waste or actually robustness - Cameron Neylon

redundancy not necessarily bad - Richard P Grant

Cameron is handing out the cyanide to the open bloggers before lunch - Jean-Claude Bradley

Issues for scientists and their careers - vulnerability to data predators, already a systemic problem in certain fields (people being sent to meetings directly to spy) - Cameron Neylon

who is affected - fields where ratio of discovery time to discovery validation in very high - hard to find but easy to show once you know - Cameron Neylon

what are the implications - if you are deprived of a paper it can be terminal - Cameron Neylon

It's good to be able to eat, even if it is cyanide... - Richard P Grant

I'm getting a bit hungry too - Jean-Claude Bradley

tweaking the rewards system - possible solutions. A short list of genes, data in a universal database, with timestampe and author ID. Could flag some to follow up and put a personal I am following this up. Do a bunch of experiments around which you then publish. Someone may ignore te request and write their own paper. WHo is the independent arbiter? Journals/entirely new body? - Cameron Neylon

Could reject paper - enforce authorship across a paper - Cameron Neylon

maybe not need papers but will still need synthesis - Richard P Grant

Move beyond the paper - what is the MPU and how is it published. The apaper as the story or the milestone. So what is the contribution if it is partial and can these be added up to provide a 'number' - Cameron Neylon

cultural shift - Richard P Grant

Nature precedings—citable - Richard P Grant

Nature Precedings - author list, doi, votes, commenting, versioning - Jean-Claude Bradley

I should add that this was Jennifer Rohn, but I don't think I can edit the original message... - Richard P Grant

I think the dichotomy/continuum is flawed—in that subscription publishing is not 'private' whereas primary data in paper notebooks are. - Richard P Grant

The problem of repeating experiments where the detail is not available - Cameron Neylon

Using freely hosted, and reliable, third party services - Cameron Neylon

re: continuum I think its a ways of presenting one aspect of the issue - but it's an agenda rather than a model - Cameron Neylon

comments/date stamp on a blog is just a matter of technology, though - Richard P Grant

Link from a description of the experiment (blog/paper/whatever) back to the experimental record - Cameron Neylon

its true its just technology but there is a lot of background in people's minds about what a 'blog' or a 'wiki' looks like. Versioning is something that has been associated with wikis and not blogs. In the end both are just php scripts running over databases - Cameron Neylon

Use of google docs to represent a further abstraction of experiments - summary and plans - Cameron Neylon

True. - Richard P Grant

using an open standard xy data representation (JCamP) allows the viewer to play with the raw data - Cameron Neylon

agree on the conflation of subscription with privacy - that's a fair point - Cameron Neylon

also the distinction between availability and accesibility as well - which is a whole other discussion - Cameron Neylon

using instrumentation as far as possible to capture what really happened versus what the human reports - Cameron Neylon

Googledocs is a nice way of doing this easily - Richard P Grant

instrument reports 'objectively' - Cameron Neylon

paper written on the wiki - still allowed (in some journals) to publish. - Cameron Neylon

References can point directly back to experimental details - even when particular analysis was done on different samples - Cameron Neylon

very powerful—linking to the experiments from the written up report, the 'paper' - Richard P Grant

...and publish it on natureprecedings - Richard P Grant

publishing in both nature precedings and Journal of Visualised Experiments http://precedings.nature.com/documen... - Cameron Neylon

-> publish the stuff that doesn't make it into papers, but is important to disseminate. - Richard P Grant

there are examples of openness in some areas leading to people getting jobs - Cameron Neylon

I for one welcome our robot overlords - Richard P Grant

instruments will talk to instruments to computers rather than humans reading the logs - Cameron Neylon

the issue of training people to do what might be seen as a new type of science - Cameron Neylon

I htink there's a danger of turning 'science' into 'technology' - Richard P Grant

agreed - but if we get the technology right we could also do more (better) science. I wonder sometimes whether I do science anymore though - Cameron Neylon

we're still talking about science - this is a drug discovery process example - Jean-Claude Bradley

Is drug discovery science? Philosophical question ;) - Richard P Grant

@Richard - in the end, doesn't science become technology and the cycle repeats? - Rajarshi Guha

drug discovery a science? Ha! More of an art :) - Rajarshi Guha

started off on a blog set up for staff by university of sydney http://blogs.usyd.edu.au - Cameron Neylon

started in August 2006 at usyd - Cameron Neylon

then moved the http://network.nature.com at 'the scientist' - Cameron Neylon

why blog? Teaching, outreach, help and advice, sharing data/reagents/protocols - Cameron Neylon

he likes writing .. nice :) - Jean-Claude Bradley

Teaching - seminar, tips, feedback from scientists, methods - still working on how to use blogging effectively for teaching - Cameron Neylon

blogging is still experimental - and that is ok - Jean-Claude Bradley

been less focussed on lab teaching but on generic stuff for scientists, how to do presentations, but also feedback from the people who are getting taught - Cameron Neylon

Outreach - talking to arts/humanities, communicating what a scientist is/does, mention of the Open Lab collection 'the eureka moment', the people who just like the pictures as art - Cameron Neylon

use blogs to get real science in real world - Jean-Claude Bradley

help and advice - worked in perl and provided advice on how to do that from a wide range of people, including computer scientists, industrial people - Cameron Neylon

Sharing - protocols, saving money (by finding the right suppliers), further help, USENET over again - Cameron Neylon

he's still concerned about getting scooped - Jean-Claude Bradley

Problems - causing offence, someone thought it was possible to identify someone anonymously. How much is allowed? What are the limitations.How much detail? Time away from the bench was a central issue, started writing only at home. A little less black and white about it now. What does the boss think? - Cameron Neylon

Doesn't have the answers at the moment - Cameron Neylon

Is any information I give actually valuable without context/reagents. Are methods more valuable than data? Need good help to good experiments with difficult methods. "Gandalf makes cells explode" - Cameron Neylon

Issues - blogging data that is presented at meetings. Is this the same thing? Stealing vs redistribution. Publication and patent rights. - Cameron Neylon

yes blogging is disclosure for patent purposes (just like publishing a paper) - Jean-Claude Bradley

Would liek to see - collaborative, more than one 'hard science' blog, central faculty commitment - support, marketing, training. Arts/humanities community don't seem to have the same worries about whether it is value or 'what the boss thinks' - Cameron Neylon

marketing teams are interested in using these things for marketing what the institution does - Cameron Neylon

mention Mat Todd and Synaptic Leap - Jean-Claude Bradley

mentioning the synaptic leap and Mat Todd's grant to do open drug synthesis development - Cameron Neylon

challening assumptions - most people have never heard of blogging, even if they don't see the point - Cameron Neylon

Challenge - is it worth it, if so how does it become mainstream - Cameron Neylon

segue into last.fm - can it be applied to research? - Cameron Neylon

http://www.mendeley.com/ - Richard P Grant

last.fm checks what you are listening to and adds that to your profile online - Cameron Neylon

lets you discover your own stats - over time, artist. Finding people who are interested in the same thing and therefore things that might be similar or of interest to people who are interested in what you are listening to - Cameron Neylon

obvious link to research papers - Cameron Neylon

combines two products - mendeley desktop manages your local papers, mendeley web runs the back end combination of data - Cameron Neylon

value without network effect - Jean-Claude Bradley

metadata extraction sounds cool - Richard P Grant

predicts metadata from text - Jean-Claude Bradley

desktop works on its own without needing the network effects - takes the paper (pdf) and extracts metadata, full text, references etc - can search tag and group and share - Cameron Neylon

can backup whole paper library - Jean-Claude Bradley

web app provides backup and access - provides profiles, social networking and groups (working on the latter) - Cameron Neylon

looking towards recommendations, research trends, and stats by author, paper, or discipline - Cameron Neylon

correction of incorrectly extracted metadata can be shared across users - Cameron Neylon

I'm impresssed - doing a live demo but the interface looks nice, does cool stuff and really seems to get a lot of this right - Cameron Neylon

working on improved work and latex integration - Cameron Neylon

this collaboration aspect gives it real power - Richard P Grant

drag and drop backup to online library - Cameron Neylon

I approve of going for the 3 major OS simultaneously - Richard P Grant

but the application works without the collaboration and already provides that added benefit by not requiring you to type in the metadata. This is a wonderful example of what I've been writing a bit about recently. Provide a good app, add a nice bit of extra functionality, and then introduce the killer next step capability once people are drawn in - Cameron Neylon

rss feeds of various aspects - Cameron Neylon

going to provide option for privacy - interesting to know what the default is because real power will come from large numbers - Cameron Neylon

I think without the collaboration it's just another—very clever, admittedly—reference manager - Richard P Grant

Don't get me wrong, I'm impressed :) - Richard P Grant

I agree its just a reference manager to start with - but that's what is needed to get people in, you need something that fits in with an existing workflow - once they are sucked in then blow them away with something new - Cameron Neylon

I might give this a shot when I get back - Richard P Grant

I've just downloaded it but I wont' try to install right now :-) - Cameron Neylon

if its reasonably bug free then I will probably get the group to use it at RAL - Cameron Neylon

you should live-blog your experience with it ... - Richard P Grant

I'm not sure I can split my attention quite that many ways at the same time - or at least i will wait until lunch in any case... - Cameron Neylon

Here's a shorter YouTube version of the talk I gave: http://www.youtube.com/watch... - Victor / Mendeley Team

I've been playing around with it and it has a lot of potential, not least because it is extremely intuitive. waiting to put more in to see if I can get more out. - Heather

Main community is sharing Taverna workflows - Cameron Neylon

Packs provide a way of collecting sets of arbitrary stuff both inside and outside of MyExperiment - provides some versioning for inside and outside objects as is possible (primarily via MD5) - Cameron Neylon

no longer just 'Facebook for Scientists' but different as well - Cameron Neylon

sharing workflows enables re-use - specific example looking at parasites in animals - Cameron Neylon

part of a distributed system - not focussing on a monolithic system - Cameron Neylon

not trying to speed up machine processing but speed up the human side - Cameron Neylon