Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

The Seven Deadly Sins of Bioinformatics

Keynote talk at Bioinformatics Open Source Conference (BOSC) Special Interest Group at the 15th Annual International Conference on Intelligent Systems for Molecular Biology (ISMB 2007) in Vienna, July 2007 by Carole Goble, University of Manchester.

  • Login to see the comments

The Seven Deadly Sins of Bioinformatics

  1. The Seven Deadly Sins of Bioinformatics Professor Carole Goble [email_address] The University of Manchester, UK The myGrid project OMII-UK
  2. Roadmap <ul><li>Sins of BioScience </li></ul><ul><ul><li>With examples </li></ul></ul><ul><li>Why are we like this? </li></ul><ul><li>The Selfish Scientist? E-Science is me-Science. </li></ul><ul><li>Challenges </li></ul><ul><ul><li>Technical </li></ul></ul><ul><ul><li>Social </li></ul></ul>
  3. Intractable Problems in Bioinformatics. Have we sinned? Are these part of the intractable problem?
  4. The traditional sins…. <ul><li>Lust </li></ul><ul><li>Gluttony </li></ul><ul><li>Greed </li></ul><ul><li>Sloth </li></ul><ul><li>Wrath </li></ul><ul><li>Envy </li></ul><ul><li>Pride </li></ul> [Stevens and Lord]
  5. Methodology <ul><li>Email a handful of bioinformaticans. </li></ul><ul><li>Stand well back. </li></ul><ul><li>Collect. </li></ul><ul><li>Edit. </li></ul><ul><li>Therapy on the cheap. </li></ul><ul><li>We all felt better. </li></ul>
  6. I am grateful to… <ul><li>Phil Lord (University of Newcastle) </li></ul><ul><li>Anil Wipat (University of Newcastle) </li></ul><ul><li>Matthew Pocock (University of Newcastle) </li></ul><ul><li>Robert Stevens (University of Manchester) </li></ul><ul><li>Paul Fisher (University of Manchester) </li></ul><ul><li>Duncan Hull (Manchester Centre for Systems Biology) </li></ul><ul><li>Norman Paton (University of Manchester) </li></ul><ul><li>Marco Roos (University of Amsterdam) </li></ul><ul><li>Rodrigo Lopez (EBI) </li></ul><ul><li>Tom Oinn (EBI) </li></ul><ul><li>Andy Law (Roslin Institute) </li></ul><ul><li>Graham Cameron (EBI) </li></ul>
  7. They came up with more than seven. But I beat them into submission. Many are highly inter-related. Hopefully they are all too familiar.
  8. Sins <ul><li>Parochialism and Insularity </li></ul><ul><li>Exceptionalism </li></ul><ul><li>Autonomy or death! </li></ul><ul><li>Vanity: Pride and Narcissism </li></ul><ul><li>Monolith Meglomania </li></ul><ul><li>Scientific method Sloth </li></ul><ul><li>Instant Gratification </li></ul>
  9. <ul><li>Parochialism </li></ul><ul><li>“ being provincial, being narrow in scope, or considering only small sections of an issue.” </li></ul><ul><li>Insularity </li></ul><ul><li>“ a person, group of people, or a community that is only concerned with their limited way of life and not at all interested in new ideas or other cultures.” </li></ul>Sin 1
  10. Reinvention <ul><li>Reinventing the Wheel. Rediscovering the same problems. Rediscovery of techniques & methods. </li></ul><ul><li>Creating… </li></ul><ul><li>Yet another identity scheme. Yet another representation mechanism for data. </li></ul><ul><li>Yet another ontology. Yet another data warehouse. </li></ul><ul><li>Yet another integration framework. Yet another query or ontology or workflow language. </li></ul><ul><li>Result? Misery. Or more work for the boys. </li></ul>
  11. Comparative Genomics? Tisk! Its Comparative Bioinformatics Bioinformatics is about mapping one schema to another, one format to another, one id scheme to another. What a waste of time. What a handy distraction from doing some Real Science™.
  12. Names and Identity Crisis Q92983 O00275 O00276 O00277 O00278 O00279 O00280 O14865 O14866 P78507 <ul><li>WSL-1 protein </li></ul><ul><li>Apoptosis-mediating receptor DR3 </li></ul><ul><li>Apoptosis-mediating receptor TRAMP </li></ul><ul><li>Death domain receptor 3 </li></ul><ul><li>WSL protein </li></ul><ul><li>Apoptosis-inducing receptor AIR </li></ul><ul><li>Apo-3 </li></ul><ul><li>Lymphocyte-associated receptor of death </li></ul><ul><li>LARD </li></ul><ul><li>GENE: Name=TNFRSF25 </li></ul>Q93038 = Tumor necrosis factor receptor superfamily member 25 precursor P78515 Q93036 Q93037 Q99722 Q99830 Q99831 Q9BY86 Q9UME0 Q9UME1 Q9UME5 Annotation history:
  13. Andy Law's Third Law <ul><li>“The number of unique identifiers assigned to an individual is never less than the number of Institutions involved in the study”... and is frequently many, many more. </li></ul>
  14. The Selfish Scientist <ul><li>“ A biologist would rather share their toothbrush than their (gene) names” </li></ul><ul><li>Mike Ashburner </li></ul><ul><li>Professor Genetics </li></ul><ul><li>University of Cambridge </li></ul><ul><li>UK </li></ul><ul><li>Amongst the many </li></ul>
  15. Some causes of the Identity Crisis <ul><li>Conflation of the ID for a thing, something to call the thing, a description of the thing, with the thing itself (reference/referent) </li></ul><ul><li>Internal vs external IDs </li></ul><ul><li>Opaque vs human-interpretable IDs </li></ul><ul><li>Situation-dependent 'parts' of a resource get different IDs </li></ul><ul><ul><li>e.g. the gene in a disease process vs the disease in a metabolic process </li></ul></ul><ul><li>Annotation attribution and log differentiation </li></ul><ul><ul><li>Two organisations attach annotations to two IDs, state they are referring to the same thing, they now have provenance about which of them asserted which facts </li></ul></ul>[Pocock]
  16. Id Reinvention <ul><li>Global Identity naming mechanism for data objects in the Life Sciences </li></ul><ul><li>LSIDs and URIs and PURLs. WS-Naming and all its friends </li></ul><ul><li>Half the debaters haven’t actually read the LSID or URL or PURL specs. Or provided use cases. </li></ul><ul><li>Web Pages are not Data Assets. </li></ul><ul><li>“ you could do this with HTTP based identifiers given <insert hack>”. </li></ul><ul><li>The debate rages! 124 messages in the last week. </li></ul><ul><li>W3C Semantic Web Health Care and Life Sciences Interest Group [email_address] </li></ul>{db}:{id}   http:// /{db }/{id}
  17. Andy Law’s First (Format) Law <ul><li>“ The first step in developing a new genetic analysis algorithm is to decide how to make the input data file format different from all pre-existing analysis data file formats.” </li></ul><ul><li>Different codes to signify the sex of animals. </li></ul><ul><li>crimap uses '0' female and '1' male. </li></ul><ul><li>Keightly algorithm. ‘1' female and ‘0' male. </li></ul><ul><li>Knott & Haley QTL analysis algorithm ‘1' female and ‘2' male </li></ul><ul><li>When they'll use '3' and '4' and then we'll know they're doing it deliberately.  </li></ul>
  18. <ul><li>EMBOSS lists more than 20 different sequence formats. </li></ul><ul><li>“ Nearly every collection of sequences that dares call itself a database has stored its data in its own format.” </li></ul><ul><li>http:// </li></ul>
  19. Reinvention of Ontology tools <ul><li>OBO and OWL ? </li></ul><ul><li>OBOEdit and Protégé-OWL ? </li></ul>The Montagues and The Capulets.. Let me get my bullet-proof vest …
  20. The “Oh No” OBO Pragmatists Aesthetics Philosophers Life Scientists Capulets Knowledge Representation Montagues A means to an end Content providers Theoreticians The end Mechanism providers Spiritual guides The Montagues and The Capulets …SOFG 2004, KCap 2005, Comparative and Functional Genomics 2004 Endurants, Perdurants, Being, Substance, Event
  21. Yet another database … <ul><li>Organism databases </li></ul><ul><li>Counter example </li></ul><ul><ul><li>Generic Model Organism Database Toolkit. </li></ul></ul>FlyBase, WormBase, SGD, BeeBase and many other large and small community databases
  22. BioBabel <ul><li>bioperl </li></ul><ul><li>biojava </li></ul><ul><li>biopython </li></ul><ul><li>bioruby </li></ul><ul><li>biophp </li></ul><ul><li>biosql </li></ul><ul><li>biouml </li></ul><ul><li>biofoo </li></ul><ul><li>biobar </li></ul>
  23. Integration <ul><li>Workflows Management Systems </li></ul><ul><li>Counter example </li></ul><ul><li>Taverna  </li></ul><ul><li> </li></ul>
  24. <ul><li>Reinvent wheels in creating 'Transcriptional Units' ('genes' derived from ESTs and mRNA), within species and between species. </li></ul><ul><li>This holds for many genome assembly related stuff </li></ul><ul><li>Genome data compilers for E. coli, Drosophila, Plant species, etcetera reuse each other's code? </li></ul><ul><li>Usually something new is added, but large parts could have been reused. </li></ul>
  25. Any more ? <ul><li>Another Web 2.0 Web Site? Another Web interface to a database? Another portal? </li></ul><ul><li>Whole database systems. ACeDB is not a lone-case. </li></ul><ul><li>Genome data compilers for E. coli, Drosophila, Plant species, etcetera reuse each other's code? </li></ul><ul><li>Text miners require synonyms and reinvent the wheel to get them in many cases. </li></ul><ul><li>Add your favourite here…. </li></ul>
  26. Reuse Rocks. Collaboration through workflow and web services <ul><li>VL-e Project </li></ul><ul><li>“ instant collaboration” with Martijn Schuemie (Rotterdam) through a web service that discloses their protein synonym data. </li></ul><ul><li>Exchanging services and (sub)workflows with food scientists. </li></ul><ul><li>Web services make that easier. </li></ul>
  27. Recycling, Reuse, Repurposing <ul><li>A Trypanosomiasis in Cattle workflow (by Paul) reused without change for Trichuris muris Infection (by Jo). </li></ul><ul><li>Identified the biological pathways believed to be involved in the ability of mice to expel the parasite. </li></ul><ul><li>Workflows are memes. Scientific commodities. To be exchanged and traded and vetted and mashed. Users add value. </li></ul>
  28. Warning! Reuse is Hard <ul><li>Writing reusable workflows is hard. </li></ul><ul><ul><li>Local services </li></ul></ul><ul><ul><li>Permissions. Licences </li></ul></ul><ul><ul><li>What does it DO? </li></ul></ul><ul><li>Writing reusable services is hard. </li></ul><ul><ul><li>What does it DO? </li></ul></ul><ul><ul><li>Predicting the unknown required by the unknown. </li></ul></ul><ul><li>Finding workflows, services and tools is hard </li></ul><ul><ul><li>Where do you go?? What does it DO?? </li></ul></ul><ul><li>Creating web services is still a bottleneck. For quick solutions it is still seen as too much extra trouble. </li></ul>
  29. Bullying and the Borg <ul><li>If a group is working in a field, you get bullied at for trying out something different. </li></ul><ul><ul><li>Can YOU think of an example?? </li></ul></ul><ul><li>You may actually be doing something different, but you use some common words. </li></ul><ul><li>“ Why do this? It's already been solved by Foo - the massively unwieldy, slow-moving, monolithic, meeting paralysed international effort for Things Mentioning Foo”. </li></ul>
  30. Reinvention or Invention? Pre-dating <ul><li>BioMOBY pre-dates (Semantic) Web service revolution </li></ul><ul><li>OBO and OBO-Edit pre-dates OWL and Protégé-OWL </li></ul><ul><ul><li>20 years of Knowledge Representation. </li></ul></ul><ul><li>Taverna pre-dates a reliable Open Source BPEL engine </li></ul><ul><ul><li>20 years of functional programming. </li></ul></ul><ul><li>There ARE features that Bioinformatics needs that other solutions don’t cater for. </li></ul>
  31. A few months in the laboratory (or the computer) can save a few hours in the library (or on Google). Westheimer's Law (with additions).
  32. No tool is an island… <ul><li>Assume </li></ul><ul><ul><li>only we will use it, whatever it may be. </li></ul></ul><ul><ul><li>that it will be freestanding and unlinked to anything else. </li></ul></ul><ul><ul><li>that it will always work and will keep on working. </li></ul></ul><ul><ul><li>That everyone will understand it. </li></ul></ul><ul><li>“Well I know what I mean. And so does my mate. So I don’t need to specify it. Or document it properly. Or keep the metadata up to date.” </li></ul><ul><li>Never mind the interface, just look at my implementation! </li></ul><ul><li>Metadata matters. Models matter. </li></ul><ul><li>Interfaces matter. Services matter. </li></ul>
  33. I know what it means... <ul><ul><ul><li>A hacker who studied ontology </li></ul></ul></ul><ul><ul><ul><li>Was famed for his sense of frivolity </li></ul></ul></ul><ul><ul><ul><li>When his program inferred </li></ul></ul></ul><ul><ul><ul><li>That Clyde ISA Bird † </li></ul></ul></ul><ul><ul><ul><li>He blamed – not his code – but zoology </li></ul></ul></ul><ul><ul><ul><li>† Clyde ISA Elephant </li></ul></ul></ul>“ AI limericks” by Henry Kautz http://
  34. Not just bioinformatics Computer Science is Guilty!
  35. Why don’t biologists modularise OWL ontologies properly? Er, well, like how should we do it “properly” and where are the tools to help us? We don’t know and we haven’t got any. But here are some vague guidelines. W3C Semantic Web for Life Sciences mailing list, 2005
  36. “ I don't blame them [MGED/PSI community] because to truly comprehend RDF/OWL is not an easy task, it takes not just the understand of technology itself but more so the vision on how things should and can work in SW.” “ One thing we have to remember is that biologists are building ontologies to do a job of work. They are not produced as some end of CS or SW research” “ Principles are all well and good, but we should know from decades of software engineering that saying &quot;do it properly&quot; isn't a solution. We need tooling and methodologies that do not in themselves hinder a domain specialist. In many cases it is easier to re-develop than re-use or even cut-and-paste from an existing ontology than it is to muck around “doing it properly”” “ There is actually a gap between the view of ontology for CS people and for biological people. The ontology in biologist's eyes are more of a treaty than logical representation, that in CS view is on the reverse of that view. It needs dialog to bring the view to a middle ground and mechanisms to stretch to both directions.”
  37. Standards are boring (but important) <ul><li>“ Blue collar Science” (John Quackenbush) </li></ul><ul><li>Nobody is going to win a Nobel prize for creating a standard schema, ontology or whatever. (Duncan Hull) </li></ul><ul><li>“ Standardise where you need standards, don’t where you don’t. Standardise messages not structures” (Graham Cameron) </li></ul><ul><li>Drive on the left or the right? </li></ul>
  38. Self promotion <ul><li>Not making shareable reusable software, because we can publish every single monolithic software solution. </li></ul><ul><li>And get promoted. </li></ul><ul><li>Applies equally to databases and ontologies. </li></ul><ul><li>Production vs Novelty </li></ul>Not all software and databases are equal.
  39. Research – Production Confusion <ul><li>Novelty vs Standards </li></ul><ul><li>Neither the funding nor the social structures of bioinformatics allow us to treat these two differently in any principled manner </li></ul><ul><li>How do you get funding for production software other than claiming to be researching stuff? </li></ul><ul><li>How do you get a publication out of a bit of research software without claiming a potential user-base? </li></ul>
  40. Trust I don’t trust your code I don’t trust your data I don’t trust you will still be around in 1 year
  41. Sin 2 <ul><li>Exceptionalism </li></ul><ul><li>Biologist exceptionalism </li></ul><ul><li>Biological exceptionalism </li></ul><ul><li>Biology exceptionalism </li></ul><ul><li>A cause of Reinvention Syndrome </li></ul><ul><li>“ Bioinformatics is special” </li></ul><ul><li>“ Domain specific outcomes requires-specific approaches and technologies” </li></ul>
  42. Biologist exceptionalism <ul><li>I know there is already a gene name for that gene, but, I don't like it and it doesn't fit in with my schema. </li></ul><ul><li>It would be better if I wrote the script I need so I know what it does, how it does it and how to modify it later because I haven’t specified what it was supposed to do in the first place. </li></ul>I’m different. We are all individuals.
  43. Biological exceptionalism <ul><li>“ Biology is all exception.” </li></ul><ul><li>“ Don’t complicate everyone’s life for the sake of a few esoteric cases”. Cameron’s 5 th Commandment of Curation </li></ul><ul><li>Exceptionalism paralysis. </li></ul><ul><li>Gather requirements expansively, prune ruthlessly </li></ul><ul><li>The EMBL/GenBank/DDBJ/Feature Table </li></ul>
  44. We are so much more complex… <ul><li>“ There are proteins, and there are records about proteins. Records come in different formats. If I make a statement using this url, is it about the record? or the protein?” Alan Ruttenberg </li></ul><ul><li>“ [Usually] we have one entry per gene. We have several entries for a single gene when description of variations are too complicated to describe in FT lines (of course, this criteria depends on the annotator). For viruses, it is much more messy, due to ribosomal frame-shifts. Formalise that!” Eric Jain UniProtDB </li></ul><ul><li>er…decomposition and untangling? </li></ul>
  45. Other Sciences…. <ul><li>CERN: UML meta-modelling mechanisms in order to migrate models over time without losing data. </li></ul><ul><li>Ensembl: “Our data models are complicated - I don't think specifying them will help. We need to understand them instead.” </li></ul><ul><li>And? </li></ul><ul><li>Confusing meta-mechanisms with models </li></ul>
  46. Biology Exceptionalism <ul><li>Biology is harder than anything else in the whole wide world because there is lots of it and its complicated. </li></ul><ul><li>Drawing graphs of data sets over time. </li></ul><ul><li>Physics wipes you off the map. </li></ul><ul><li>The real problem is complexity not scale. </li></ul><ul><li>The number of data sets, their diversity and how they overlap. </li></ul><ul><li>How they change. </li></ul><ul><li>Their Reliability. </li></ul>
  47. Sin 3 <ul><li>Autonomy or death! </li></ul><ul><li>Combined with churn and indifference to users. </li></ul><ul><li>Compounded by the Early Adopter tendency of the community and a monopoly mentality. </li></ul><ul><li>“ Hell is other people’s systems” as John Paul Sartre would have said if he had been a bioinformatician. </li></ul>
  48. Autonomy is death! <ul><li>Change my interface / format whenever I feel like it, despite the fact I wanted lots of users and I have lots of users who depend on this. And I won’t bother to debug either or provide backwards compatibility. </li></ul><ul><ul><li>BioMART changed 4 times in the past year. </li></ul></ul><ul><ul><li>NCBI changes as it fancies. </li></ul></ul><ul><ul><li>Ensembl relational schema. </li></ul></ul><ul><ul><li>Early BioJava. </li></ul></ul><ul><li>This is just unprofessional. </li></ul><ul><li>Stable Metadata matters. Stable Models matter. Stable Interfaces matter. Stable Services matter. </li></ul>
  49. Lincoln Stein said a while ago… <ul><li>An interface is a contract between data provider and data consumer </li></ul><ul><li>Document interface; warn if it is unstable </li></ul><ul><li>Do not make changes lightly </li></ul><ul><ul><li>Even little fiddly changes can break things </li></ul></ul><ul><ul><li>Provide plenty of advance warning </li></ul></ul><ul><li>When possible, maintain legacy interfaces until clients can port their scripts </li></ul><ul><li>Support as many interfaces as you can </li></ul><ul><li>HTML (least desired) </li></ul><ul><li>Text only (better) </li></ul><ul><li>HTTP-XML (even better) </li></ul><ul><li>SOAP-XML (sweet!) </li></ul><ul><li>Easy Interfaces + Power User Interfaces </li></ul>… and he could say it again today.
  50. Law's Second Law <ul><li>“Error messages should never be provided” corollary... “If error messages are provided, they should be utterly cryptic so as to convey as little information as possible to the end user” </li></ul>
  51. Workflow commodities <ul><li>Workflow published with its paper and its data set. </li></ul><ul><li>So what happens when I want to run this workflow again? </li></ul><ul><li>Is the service dead? </li></ul><ul><li>Is the dataset still there? </li></ul><ul><li>Was it designed to be reproduced or reused in the first place? </li></ul>
  52. The myGrid Semantic Sweatshop <ul><li>Services and Workflows in the wild. </li></ul><ul><li>Curated by experts using an ontology. </li></ul><ul><ul><li>Supplied by service providers (like EMBOSS) in text. </li></ul></ul><ul><ul><li>Or annotations (like BioMOBY, but they aren’t good annotations!) </li></ul></ul><ul><ul><li>Tagged by the Masses. </li></ul></ul><ul><li>Multi-perspective </li></ul><ul><ul><li>Scientist for finding. </li></ul></ul><ul><ul><li>Machinery for validation. </li></ul></ul><ul><li>Hard work. Look how tired they are. </li></ul>Semantic
  53. The myGrid Semantic Sweatshop notice how tired they look Franck Tanoh Katy Wolstencroft
  54. Churn, Churn, Churn <ul><li>“ Stability is more important than Standards or Smartness. Discuss” </li></ul><ul><li>Constant churn and change for change sake. </li></ul><ul><ul><li>Impact on everyone else who uses the previous mechanism. </li></ul></ul><ul><li>A few voices, very loud, vested interest, for their application, win. </li></ul><ul><li>You know what? Why don’t we stick with something for a while and rally behind it? Or at least figure out the cost of change. </li></ul><ul><li>Maybe this is a sin inherited from Computer Science. </li></ul>
  55. Churn, Churn, Churn <ul><li>We expect the content to change, but why does everything else. </li></ul><ul><li>Constant churn and change for change sake. </li></ul><ul><li>Maybe this is a sin inherited from Computer Science. </li></ul><ul><li>The W3C Identity War. Web Services vs REST </li></ul><ul><li>Impact on everyone else who uses the previous mechanism. </li></ul><ul><li>A few voices, very loud, vested interest, for their application, win. </li></ul><ul><li>You know what? Why don’t we stick with something for a while and rally behind it? Or at least figure out the cost of change. </li></ul><ul><li>“ Stability is more important than Standards or Smartness. Discuss” </li></ul>
  56. Sin 4 <ul><li>Vanity </li></ul><ul><li>Pride </li></ul><ul><li>Narcissism </li></ul><ul><li>conceit, egotism or simple selfishness. </li></ul><ul><li>Applied to a social group, denotes elitism or an indifference to the plight of others </li></ul>
  57. I know it all. <ul><li>Claiming to know everything about biology and everything about computers. </li></ul><ul><li>This is really irritating to both biologists and computer scientists. </li></ul><ul><li>Even they don’t claim to know everything about biology or computer science. </li></ul><ul><li>Computer scientists do know a lot of stuff. And they publish too. </li></ul><ul><li>“ Biologists are the experts on everything because we produce the data” </li></ul>And what would you suggest, Mr. Smartie Pants?
  58. Think like me! <ul><li>Building interfaces that only you can use. </li></ul><ul><li>Not actually using your tools in the field. </li></ul><ul><li>I understand workflows </li></ul><ul><li>Workflows are for biologists. </li></ul><ul><li>My granny can do workflows... </li></ul><ul><li>Designing good experiments is hard. </li></ul><ul><li>Workflows are computational experimental protocols. Ergo…. </li></ul><ul><li>Writing workflows should be expected to be hard. </li></ul><ul><li>Writing good workflows is really hard. </li></ul><ul><li>Writing good reusable workflows is really really hard. </li></ul>Misunderstanding and disrespecting users
  59. A good User Experience outweighs smart features. Can I use it? Is the user interface familiar? Does it fit with my needs?
  60. Gain-Pain pay-off <ul><li>Just enough, just in time </li></ul>Gain Pain Very BAD Good, but Unlikely Just right
  61. Sin 5 <ul><li>Monolith Meglomania </li></ul><ul><li>delusions of grandeur. </li></ul><ul><li>obsession with grandiosity and extravagance. </li></ul><ul><li>Data mining - “my data is mine, and your data is mine” </li></ul>
  62. More, more, more! <ul><li>Integration – the more the merrier. No. </li></ul><ul><ul><li>Every link is a potential dead link. </li></ul></ul><ul><ul><li>Every dependency can find its way on to your critical path. </li></ul></ul><ul><ul><li>Monolithic solutions always fail. </li></ul></ul><ul><li>Put it all in a warehouse. </li></ul><ul><ul><li>ATLAS, MRS, e-Fungi, GIMS, Medicel Integrator, MIPS, BioMART blah blah blah… </li></ul></ul><ul><ul><li>Toolkits: Information Integrator, GMOD, BioMART, BioWarehouse, blah blah… </li></ul></ul><ul><ul><li>50% warehouses fail. </li></ul></ul><ul><li>Uber-tools” and “Uber-databases” </li></ul><ul><ul><li>Biomart, Ensembl, etc etc…. </li></ul></ul>[Cameron]
  63. The trouble with warehouses <ul><li>30% of data migration projects fail (Source: Standish Group) </li></ul><ul><li>50% of data warehousing / Business Intelligence projects fail (Source: NCR) </li></ul><ul><li>“ Warehouses work? Piffle. They never manage to maintain synchrony with the source data. Mostly they fall down of their own weight!” Graham Cameron, EMBL-EBI </li></ul><ul><li>&quot;Our ability to capture and store data far outpaces our ability to process and exploit it. This growing challenge has produced a phenomenon we call the data tombs, or data stores that are effectively write-only; data is deposited to merely rest in peace, since in all likelihood it will never be accessed again. Data tombs also represent missed opportunities.&quot; Usamma Fayyad Yahoo! Research! Laboratories! </li></ul><ul><li>We believe that attempts to solve the issues of scientific data management by building large, centralised, archival repositories are both dangerous and unworkable” Microsoft 2020 Science report. </li></ul>
  64. More More More <ul><li>“ Emacs of Biology” </li></ul><ul><li>End-user apps/libraries in bioinformatics workbenches with loads of crap bundled in, none of it kept up to date, none of it properly integrated. </li></ul><ul><li>Keep it simple and modular </li></ul><ul><li>Don’t reinvent Eclipse. </li></ul>
  65. Mash-Up Data Marshalling <ul><li>Content syndication and feeds </li></ul><ul><li>Emphasis shifts to the user creating specific integration by mapping. </li></ul><ul><li>Just in time, just enough design </li></ul><ul><li>On demand integration – or rather, aggregation. </li></ul>Mash Up Application User interface Protocol objects Protocol Protocol
  66. Distributed Annotation System Mash-Up Reference Server AC003027 AC005122 M10154 Annotation Server Annotation Server AC003027 M10154 WI1029 AFM820 AFM1126 WI443 AC005122 Annotation Server
  67. Sin 6 <ul><li>Scientific Method Sloth </li></ul><ul><li>Its easier to think of a new name than use someone else’s. </li></ul><ul><li>I want my own view over data and views are difficult, so I’ll create my own database. </li></ul><ul><li>Leads to Reinvention, Exceptionalism </li></ul><ul><li>Often the result of Instant Gratification </li></ul>
  68. Ennui <ul><li>Garbage in, garbage out </li></ul><ul><ul><li>Running analysis over the wrong datasets </li></ul></ul><ul><ul><li>E.g. Identifying chicken proteins in mouse cells. </li></ul></ul><ul><li>Configuration traditionalism </li></ul><ul><ul><li>Not changing the parameters of BLAST. Ever. </li></ul></ul><ul><li>Top list ennui </li></ul><ul><ul><li>If there is a list only looking at the first one. </li></ul></ul><ul><ul><li>Look no further than the first Blast hit / first Google hit. </li></ul></ul><ul><li>Arbitrary cut-offs on rank-ordered result list </li></ul><ul><ul><li>Absolute truth above, absolute falsehood below </li></ul></ul><ul><ul><li>E.g. differentially expressed genes in microarray analyses. </li></ul></ul>
  69. Its black and white <ul><li>Arbitrary cut-offs on rank-ordered result list </li></ul><ul><li>Everything above is absolute truth and everything below complete falsehood.  </li></ul><ul><ul><li>sequence similarity when looking for orthologs. </li></ul></ul><ul><ul><li>protein identifications using Mascot scores. </li></ul></ul><ul><ul><li>differentially expressed genes in microarray analyses. </li></ul></ul>
  70. Quality Delusions <ul><li>The bioinformatics does not have to be sound, because we only trust wet-lab results anyway. </li></ul><ul><li>Worrying about errors in experimental data but believing that derived data is always true. </li></ul><ul><li>Believing Trembl is always right. </li></ul><ul><li>Believing computational gene predictions are always correct. </li></ul>
  71. Quality Delusions <ul><li>The bioinformatics does not have to be sound, because we only trust wet-lab results anyway. </li></ul><ul><li>Worrying about errors in experimental data but believing that derived data is always true. </li></ul><ul><li>Believing Trembl is always right. </li></ul><ul><li>Believing computational gene predictions are always correct. </li></ul>
  72. Black Box Science <ul><li>Producing irreproducible bioinformatics analyses </li></ul><ul><ul><li>Not collecting the provenance of the analysis. </li></ul></ul><ul><ul><li>Not testing during software development. </li></ul></ul><ul><li>Try re-running experiments described in the journal Bioinformatics from before 5 years ago </li></ul><ul><li>UniGene </li></ul><ul><ul><li>What is happening during UniGene clustering? </li></ul></ul><ul><ul><li>‘ Human’ descriptions (via NCBI), are not exact. </li></ul></ul><ul><ul><li>The Human Transcriptome Map project and other microarray analysts ended up reclustering UniGene [Marco Roos]. </li></ul></ul>
  73. “ No experiment is reproducible.” Wyszowski's Law “ An experiment is reproducible until another laboratory tries to repeat it.” Alexander Kohn
  74. Sin 7 <ul><li>Instant Gratification </li></ul><ul><li>Greed? Gluttony? </li></ul><ul><li>Always the immediate return. </li></ul><ul><li>Never investing for the future. </li></ul><ul><li>The quick and dirty fix. </li></ul><ul><li>Refusing to model or abstract. </li></ul><ul><li>Refusing to plan for recording and exchanging. </li></ul><ul><li>Just getting the next quick fix. </li></ul><ul><li>The pressure to deliver now and pay later </li></ul> .
  75. Hackery <ul><li>Deliver now, pay later </li></ul><ul><ul><li>Producing crap, non-reusable, software because only the biological results matter for publication X. </li></ul></ul><ul><ul><li>Collect! Analyse! Er…now what? </li></ul></ul><ul><li>Spaghetti-ism </li></ul><ul><ul><li>Over-indulgence in PERL </li></ul></ul><ul><ul><li>Over-indulgence in Ascii Art flat files. </li></ul></ul><ul><ul><li>Modelling a system by hacking up XSD fragments on a whiteboard. </li></ul></ul><ul><ul><li>Writing perl scripts that resemble my high-school BASIC of the 80s. </li></ul></ul>
  76. “ I am sure one could reuse large parts of re-annotation for building transcriptome maps, if they only used workflows and ontologies”. Marco Roos A Biologist and Bioinformatician VL-e Project, Amsterdam
  77. “ Bioinformaticians have reached the standards of the 1980s, while computer scientists are working on the standards of the 2020s, leaving roughly 40 years to bridge. Marco Roos A Biologist and Bioinformatician VL-e Project, Amsterdam
  78. Blind faith in XML <ul><li>It’s in XML, thus all data integration problems are solved. </li></ul><ul><ul><li>Er…no. </li></ul></ul><ul><ul><li>All those vocabularies e.g. SBML, GenBank XML etc </li></ul></ul><ul><li>The good thing about XML is that it is human readable. </li></ul><ul><ul><li>Arrrrgh! </li></ul></ul><ul><li>Insisting that XML is not text. </li></ul><ul><li>Insisting that XML is text </li></ul>XML
  79. Blind Faith in Foo. <ul><li>There's a new thing to use. </li></ul><ul><li>we don't understand it yet. </li></ul><ul><li>so it sucks up all the stuff we already know we don't understand. </li></ul><ul><li>Lack of appreciation about exactly what the new technology addresses in itself before trying to make it work for us . </li></ul>
  80. Pioneering development methods <ul><li>Development by anecdote </li></ul><ul><ul><li>I heard in the pub that the way to go was Foo. </li></ul></ul><ul><ul><li>Though I have no idea what Foo is or why it is the way to go. </li></ul></ul><ul><li>Design by hacking </li></ul><ul><ul><li>It would be better if I wrote the script I need so I know what it does, how it does it and how to modify it later because I haven’t specified what it was supposed to do in the first place. </li></ul></ul><ul><ul><li>Hmmm…..We call that Extreme Programming or Emergent Semantics or Web 2.0 in CS  . </li></ul></ul>
  81. Open Source Blinkers <ul><li>Why does Open source have special merit? </li></ul><ul><li>Commercial solutions with added special sauce can rock too. </li></ul><ul><li>Shall I duck? </li></ul>
  82. Sin Summary Maybe only one “original sin” in bioinformatics. Parochialism and Insularity Exceptionalism Autonomy or death! Vanity: Pride and Narcissism Monolith Meglomania Scientific method Sloth Instant Gratification Reinvention Churn
  83. Can we become less sinful? Why do these sins exist? Are bioinformaticians particularly naughty? No naughtier than Computer Scientists. And its all very hard. Though they are naughty…
  84. Why? <ul><li>Selfish Scientist – Self-interested Scientist </li></ul><ul><ul><li>Reputation, need to get results right now, win. </li></ul></ul><ul><ul><li>Fear of dependency, fear of being left behind. </li></ul></ul><ul><ul><li>Understand the incentives and barriers to adoption. </li></ul></ul><ul><li>Bioinformatics as it is practiced </li></ul><ul><ul><li>Social and funding structure perpetuates this. </li></ul></ul><ul><ul><li>Production vs Research. </li></ul></ul><ul><li>Real, inherent issues. It is hard. </li></ul><ul><li>Hybrid exhaustion and pressure. </li></ul><ul><ul><li>Biology + Computing + Bioinformatics </li></ul></ul>
  85. Luddism? Surely not! <ul><li>Refusing to have biology go beyond a cottage industry. </li></ul><ul><li>Being scared to do it properly. </li></ul><ul><li>Railing against big science </li></ul><ul><li>The cult of amateurism. </li></ul>[Stevens]
  86. Research – Production Confusion <ul><li>Novelty vs Standards </li></ul><ul><li>Neither the funding nor the social structures of bioinformatics allow us to treat these two differently in any principled manner </li></ul><ul><li>How do you get funding for production software other than claiming to be researching stuff? </li></ul><ul><li>How do you get a publication out of a bit of research software without claiming a potential user-base? </li></ul>
  87. Practical Steps? <ul><li>Create means to share know-how </li></ul><ul><ul><li>Understanding outside my expertise. e.g. sources of error. </li></ul></ul><ul><ul><li>A comprehensive catalogue of web services </li></ul></ul><ul><ul><li>A Facebook for workflow builders. </li></ul></ul><ul><ul><li>Learn from others. Even Computer Science. And other Sciences. </li></ul></ul><ul><ul><li>Try and create a culture of raising quality. Somehow. </li></ul></ul>
  88. FaceBook & Bazaar for Workflow e-Scientists Trials start August 2007!
  89. Delivery Bulge
  90. Practical Steps for IT Platforms? <ul><li>Stop building monolithic solutions </li></ul><ul><ul><li>Strong force in business enterprises </li></ul></ul><ul><li>Component-ise Bioinformatics </li></ul><ul><ul><li>Loosely coupled systems </li></ul></ul><ul><ul><li>Stable APIs, standardised metadata. </li></ul></ul><ul><ul><li>Design to combine. </li></ul></ul><ul><ul><li>Sort out the b***dy naming/id problem </li></ul></ul><ul><ul><li>If you can’t agree, agree on the bridge. </li></ul></ul><ul><li>Raise the level of abstraction </li></ul><ul><ul><li>Less Perl, more workflows  </li></ul></ul><ul><ul><li>Enable users to extract the data they need without hassling you. </li></ul></ul>
  91. Practical Steps? <ul><li>Presume and design for incremental change </li></ul><ul><ul><li>Minimise disruption. </li></ul></ul><ul><li>Presume others use our stuff </li></ul><ul><ul><li>And respect that </li></ul></ul><ul><ul><li>Describe to build Trust </li></ul></ul><ul><li>Presume others add value to our stuff </li></ul><ul><ul><li>Be easily part of loosely coupled systems. Lightweight programming models. </li></ul></ul><ul><ul><li>Presume, and enable, content and function mashing. </li></ul></ul>
  92. Web 2.0 Design Patterns <ul><li> </li></ul>26/2/2007 | myExperiment | Slide <ul><li>The Long Tail </li></ul><ul><li>Data is the Next Intel Inside </li></ul><ul><li>Users Add Value </li></ul><ul><li>Network Effects by Default </li></ul><ul><li>Some Rights Reserved </li></ul><ul><li>The Perpetual Beta </li></ul><ul><li>Cooperate, Don't Control </li></ul><ul><li>Software Above the Level of a Single Device </li></ul>
  93. Practical Steps? <ul><li>Presume scientific practice naughtiness </li></ul><ul><ul><li>Try to deal with it, or expose it? </li></ul></ul><ul><ul><li>Transparency and accurate collection and reporting. </li></ul></ul><ul><ul><li>Provenance. </li></ul></ul><ul><ul><li>A prerequisite to publication. </li></ul></ul><ul><ul><li>The end of Black Box Science. </li></ul></ul><ul><ul><li>Peer pressure. </li></ul></ul><ul><ul><li>E.g. Workflows, but will a scientist give away their secrets or expose their mistakes? </li></ul></ul>
  94. The Final Word Sin writes histories, goodness is silent.   Thomas Fuller