Mo ichido

4 years after starting at the DTC the fateful day everybody has in the back of their mind has arrived.  The viva is scheduled for 2pm on Thursday.  I’ve read the thesis a dozen times over Christmas, read papers, and chewed over the “viva survivor” websites.  I’ve got a list of corrections to submit on the way in, which might negate the horrendous pile of typo’s that were left in the insane rush to submit on time.  I’m still not sure how to answer some of the more generic questions with politically correct answers, but I’ll ponder some more tomorrow.

I didn’t get any publications out of the work so I’m expecting a rough time from the examiner.  Personally, I believe that no publications shows there was some serious problems, and I would question no contribution to the field, which is the essence of a Ph.D.  However, sometimes things just don’t go well.  It’s science.  I did what I could within the boundaries of the available resource and time.

It’s been a crazy ride in academia over the past few years.  I’ve had some pretty dark days but have met some truly excellent people along the way, and attended some excellent conferences.  I got to engage in scientific debate with people from all over the world, and learned a lot from everybody I met.  I certainly wouldn’t recommend a Ph.D to people who don’t enjoy inflicting damage on themselves.  I guess I would sum up the essence of it by paraphrasing JFK’s Rice University speech:

“We choose to go to the moon in this decade and do the other things, not only because they are easy, but because they are hard, because that goal will serve to organize and measure the best of our energies and skills, because that challenge is one that we are willing to accept, one we are unwilling to postpone, and one which we intend to win.”

I’ll try and continue posting on the blog, irrespective of the result.  I will have more time to write about the stuff I’m reading, and you never know, perhaps even continue in the field!  I would like to get more into Matlab and modelling what I was doing, so I might be able to post some dry stuff.

Anyway, back to the thesis.  Cheers to all who commented on the blog over the years.

See you on the flip side.

dynamics of disease – day 3

I missed day 2 due to travel costs.  Day 3 kicked off with Professor Hans Westerhoff presenting his vision of the systems biology of disease and 7 billion virtual humans.  Present day diseases are molecular diseases, incorporating multiple genes and therefore nodes in the network.  Modern diseases are therefore “network diseases”.  Treatments should therefore target multiple nodes of the network and not the individual molecule.  Professor Westerhoff presented his vision for linking global networks of data to provide diagnostic tools on individuals for personalized medicine.  There was then an outline of his concepts of network fragility, which looked quite similar to network sensitivity to me, but provide a mathematical prediction of intervention points in the network, based on the fragility of the network around that node, or hub with regard to the flux through the network.

Thomas Eissing, from Bayer next gave a very informative talk on the application of multi-scale modelling in medicine and pharmacology.  His group work in a consultancy role for pharma customers and are able to make multi-scale models of diseases, using pharmacokinetics (PK) (what the molecule does) and pharmacodynamics (PD) (what the molecule does to the body) to reduce clinical trials group sizes, and target more refined network intervention points for disease therapies.  Their group has developed a number of modelling tools to assist them, such as PK.Sim, and Mobi, with network topology and flux balance analysis to build static models of the “virtual human” which can be refined into dynamical models as hypothesis generation leads to experimental data.

This was the most interesting talk for me, with real-world practical applications to some of the theories outlined by Professor Westerhoff.  The group has been able to build multi-scale virtual liver models for studying tumour development, incorporating signal transduction models into cell models, and then into organ scale models with “virtual lab rats”.  Superimposing PK data onto these models then provides dynamic time course simulations to estimate drug efficacy, and also the required treatment time which can dramatically reduce the time scale of clinical trials.

Later in the day, Yusur Al-Nuaimi, one of the doctoral training centre students presented her work on systems biology approaches to human hair cycle.  Yusur had gained her Ph.D the previous day, and had recently published her work in this field (pubmed ID’s 20590819 and 19725870).  A mathematical approach firstly identified a delayed feedback loop in the life cycle of the hair follicle, and this was explored experimentally with clock genes for anagen and catagen transition, revealing an autonomous timing mechanism in the hair follicle.

The meeting finished with a presentation from Roger Traub (IBM) on the cellular mechanism of epilepsy, showing how modelling approaches can be used to try and understand the complex high frequency, low amplitude (200Hz) oscillations in brain activity that leads to seizures.  Roger Traub also used a cellular automaton model to show coordinated oscillations in populations of cells during seizures.

The Dynamics of Disease – University of Manchester

Continuing to get myself out and about during unemployment, I managed to get on to the dynamics of disease workshop at University of Manchester this week.

The conference covers the application of systems biology and systems approaches to studying disease, and is being organised by Professor Gerold Baier of the Manchester Interdisciplinary Biocentre doctoral training centre.  Gerold has provided a lot of support to me during my Ph.D, particularly in the first year, and I am very grateful that he allowed me to attend the event.

Day 1  was entitled “Systems approaches to Health and Disease” and provided a general introduction to how systems biology is being used in academic research into diseases, as well as pharmaceutical drug discovery programs.

Pedro Mendes of University of Manchester presented a general introduction to modeling biochemical networks, and described the process of moving from annotated genome sequence data to dynamic – models, with metabolic network reconstruction.  The process of building models was described, using Yeast 5 as an example “big model”.  There was some examples of analysis using FBA and MCA to interrogate these models as “static models” for hypothesis generation, and then build rate laws to create dynamic models.

Later on, Jeurgen Pahle gave a talk on stochastic modeling, and demonstrated stochastic modeling using Copasi, and the benefits it provides in observing non-linear dynamic behaviors in models, using the Lotke-Voltaire model as an example, where discreet events in stochastic simulations lead to extinction, which are not observed from averaging in deterministic modeling.

Mike White presented his groups work on NF-Kappa B signaling in cancer, and how a combined experimental and modeling approach has lead to the identification of feedback loops in the network and an understanding of the network architecture.  The modeling work provides hypothesis on how signaling functions as an oscillation in NF-Kappa B activity, and regulation by TNF-alpha can change the amplitude and periodicity of the oscillations, leading to heterogeneity in the signaling response of the population.  The work shows the real benefit of a systems level approach, and understanding the interaction of the components of the network at the cellular level, as a network embedded in many other networks, and also in a cell population level, with cells signaling to each other as a tissue, or an organ and how each cell in a component of a larger system.

Jonathan Swinton presented his work modeling cardiac action potentials for Astra Zeneca’s drug discovery programs.  Again, it was demonstrated how basic mechanistic models can provide a frame work for understanding a system, and then parameter estimation using experimental data can refine the model behavior, providing in vivo relevant simulation tools.  The library size for drug screening programs can be significantly reduced by investigating systems level interactions with the computer, leading to cheaper and faster drug discovery in the future.

In the afternoon, Ed Kent (Mendes group) provided a demonstration of sensitivity analysis and how it can be used to infer control points in metabolic networks, and his web interface for Copasi to run simulations on the Condor distributed computing network to reduce years of computation time to hours, enabling global sensitivity analysis in increasingly large models.

In addition to the speakers, there was a skype call to Joseph Loscalzo of Harvard Medical School, who is applying systems biology as “systems physiology”, attempting to understand the complexity of disease beyond simple cause and effect and Occams Razor, to understanding complex combinations of symptoms, and links between different diseases, such as Crohns disease and type II diabetes, using genome wide association studies, creating what he calls a “Pathophenome”.  He also highlighted the complexity of targeting diseases such as cancer with single drug therapies, where knocking out one interaction in the pathway can be compensated for further down the pathway, or in connected pathways, resulting in the disease returning.  Cheaper whole genome sequencing in the future will possibly enable clinicians to identify multi-point purturbations with combinations of drugs, personalised for the individual disease case.

It was great to be back among the scientific community again.  I couldn’t make day 2 due to travel costs, but I will be headed back for day 3 for “multi-scale modeling and systems dermatology”.

Simbiology webinar

This week I attended the mathworks webinar on Simbiology.  I had seen this software used by researchers at Imperial Collage and was interested to see what it could do in comparison with Copasi.  The webinar was delivered using Webex and was well presented with a mixture of slides to structure the talk, and live matlab interaction.

Simbiology is an extension to matlab and provides a point and click style interface for building models of interactions, performing simulations, and outputting graphical data.

Simbiology can import data from a number of databases, such as Kegg, and can import entire pathways into a graphical tool, similar to JDesigner.

Reactions are inputted using a similar interface to Copasi, which includes species, reactions, and compartments, as well as the facility to code custom rate laws.

Simulations can be run using deterministic, stochastic, and hybrid integrators.  Simbiology can also perform parameter estimation using experimental data to fit the model behavior.

Overall, Copasi appears to incorporate a larger number of parameter estimation algorithms, and this remains the major strength of it over the likes of Simbiology.  Simbiology however integrates with the command line of matlab, enabling users to drop into the code at any time and write/edit their own functions.  Simbiology can interface with many different databases to pull down model parameter data from virtually anywhere, and provides a much more flexible tool for modeling than Copasi.  The user is also presented with the opportunity of coding additional functions and tools from Matlab, which is not possible in Copasi without editing the applications source code, in C.

Simbiology can perform sensitivity analysis, but appears to lack metabolic control analysis (although the two are synonymous, so I don’t suppose this really matters).  Simbiology is also unable to analyse steady states, and expand around them to find multiple steady states and bifurcation points.  Bifurcation analysis is apparently being worked on for future versions, but no further information was available on the development of this.  I was surprised that once again XPPAUT appears to retain it’s unique ability to do this from a GUI, and remains the best software for this kind of analysis.

For the price difference, Copasi contains a significant proportion of functions of simbiology, and is certainly sufficient for systems biologists, with the language bindings to access it from the command line. Coupled with XPPAUT, these tools provide more features than simbiology.

Simbiology is going to continue to develop however, and the flexibility of matlab, together with the awesome amount of code available at the matlab file exchange provides a much more flexible toolkit for the hardcore modeler.  Simbiology could also potentially output graphs using matlab plotting functions, which would enable surface plots, and matlabs extensive graphics functions, which is lacking from Copasi.  Statistical analysis could also be done alongside the simulations as well, which would require additional software if using other open source applications.  For the freelancer or poor academic however, the open source tools still give mathworks a run for their money, for now.  If you have a matlab + simbiology license already however, it’s a comprehensive tool set that is accessible to biologists, and builds a matlab coding skill alongside modeling the biology.

Bioconference Live 2011

I’ve spent the last couple of days at Bioconference Live 2011.  What’s unique about this conference is it’s hosted entirely online!  The conference covers subjects across life sciences including neuroscience, drug discovery, cancer Research, genetics, microbiology, and cell biology.  There have been a number of speakers from all over the world, delivering 30 minute presentations from their offices and labs.  Many of the talks were around diagnostics and pharmaceutical research, and were a bit outside of my area of interest, however there were some interesting talks on next generation sequencing technologies and personalized medicine that were very informative.  The conference was running EST so it was a bit late for me on GMT, but I was able to attend a few of the talks before I passed out!   I particularly enjoyed the talk by BlueSEQ on next generation sequencing.  The company offers a freelance review of current sequencing technologies, and consult on the technology for researchers.  Rory McNeil also gave a talk on mobile lab applications, and eCat from Axiope, which looked like an interesting sample management/electronic notebook tool that can be used with with desktops as well as mobiles and tablets.

The technology driving the conference enabled speakers to give a live video stream of themselves, next to their presentation screen.  Questions could be posted through the interface, and the speakers could respond after their presentation, and made each talk interactive with the audience.  Talks could also be imported into calendar tools so you could pick and chose your talks throughout the day, whilst carrying on with other work.  Slides could also be added to a virtual briefcase and downloaded after the talk.

Everybody likes conferences for freebies, and there was also an exhibitor hall.  The hall contained a virtual mall of exhibitor stands that could be scrolled through, and then clicked on for a particular vendor.  Each stand then had a multimedia presentation, links to produces and information, and free offers of samples.  Vendors could also be contacted live via video link to ask questions and get more information.

Also, there was a networking lobby using a chat room, where conference visitors could meet and chat with each other.  There were reps and exhibitors, as well as delegates and it was an excellent facility for meeting new people.

There were a diverse range of keynote speakers, listed here covering a whole range of different fields and specialisms.

Overall, the conference was excellent, and provides people like myself, no longer working in the field, the opportunity to keep on attending scientific conferences.  In the current financial climate, the technology shows it’s possible to host big international conferences with no travel costs, and retain many of the features of a live event, that can be fitted around the increasingly busy working schedule.  Hopefully it is a sign of things to come, rather than an exception.

There are further conferences from the organizers, and there appears to be another life sciences event in September 12-13 2012.

Back to the talks!

11th hour

They do say if it wasn’t for the 11th hour nothing would ever get done.  72 hours to submission and I get a re-write dropped on me, oh and some notice that I should expect the worst from the viva as everything I’ve done is rubbish.  Well, the latter part I didn’t need to be a professor to know (shame I wasn’t rubbish at saving the lab £5,000 when I fixed the AKTA after GE gave up, or when I did those stochastic simulations for that grant proposal, or the numerous rotation projects I magicked up, or the RT-qPCR machine I got for half price, not to mention moving the lab across the country, ho hum.).  30 hours without rest has resulted in most of it implemented, but the discussion chapter still needs serious re-writing, and all the figures re-drawing.  An emergency appeal to the university has resulted in a 7 day extension but with a late submission mark against my name.  Not too bothered about that, at this point.  A last minute reprieve or just delaying the inevitable?  I don’t know.  I get a few more hours added to my clock to pluck some more magic out of the air.  It wont be corrected, and it wont be pretty.

At least the project has remained consistent to the end.

I’ll be glad to be out of academia, one way or the other.

all stop

Friday will be the last day in the lab (theoretically).  It’s tools down and writing time.  I’ve got about 4 weeks maximum to get the thesis written and this nightmare ended.  The last several weeks have been a blur of all night luciferase measurements, western blots, northern blots, and qPCR.  For now, life is a blur of LaTeX.

There’s precious little data and the models are questionable to say the least. I will throw it all out to the examiner and let them make the decision.  I’ve flogged this dead horse until it’s dust.

Facebook account is disabled until October.  All appointments are cancelled.  I’m hoping to not have to do the road trip down to the lab and stay at home, but the boss probably wont go for that.  It’s the usual curse of the wet lab scientist – unless you’re in a lab coat you’re not working, and that includes writing.

I hope I can get this done in time :/

See you on the flip side.

the one man band

I felt compelled to write a post after reading an article in this month’s Nature Careers entitled “Seeking the right toolkit” (Nature 7358:476).  The article attempts to be balanced, but plays heavily on the benefits of inter-disciplinary training.

I believe there is a miss-understanding in the terminology between “inter” and “intra” disciplinary, and I don’t think it’s merely pedantics.  I believe that it underpins a fundamental problem in the training programs.

I whole heartedly believe that inter-disciplinary research is the future of science.  Mathematics, through systems biology is now pushing into the biological sciences, as bioinformatics has been doing for a number of years.  Studies of single molecule interactions has led to the discovery of complex networks of interactions, and the implementation of high throughput automation has increased the number of parameters that can be investigated simulataneously, as well as the number of technical replicates.  Technological improvements in computer technology has fueled emerging fields such as systems biology that attempt to go beyond studying single interactions to re-constructing entire pathways at various hierarchical levels within the cell.  Consequently, the understanding of these systems is so complex that interpretation of the data requires mathematical tools for simulation and hypothesis generation.   Emerging from this evolution are training programs designed to produce a next generation of scientists that can understand these diverse component fields and employ all of the available tools from them in their future research.

Yet there is a paradox between the programs and the projects.  I believe the training programs are attempting to create their own requirements.  The doctoral training centers aim to produce intra-disciplinary students through a year of training in both mathematical and biological sciences, combining these skills into a single individual.  However, the research projects running in the host centres are still focused into various specialist labs.  The doctoral training centre provides the opportunity for these labs to branch out into inter-disciplinary research through the proposal of projects that the students can apply for. Consequently, there can be no experience or history of inter-disciplinary research in the host lab and the student is expected to both be an expert from the 1st year training, and fullfill all of the roles that would otherwise be provided by individual specialists coordinating on a multi-disciplinary project.  Intra-disciplinary skill sets in the students are not being used as a next-generation research approach, but out of necessity to plug the gap in a lack of resourcing.  More importantly, poorly planned projects don’t generate high impact papers, or any results at all.  While publications are enjoyed by the lab, any failure is lumped onto the student’s incompetence and the principle investigator can pass the responsibility back to the training centre.  If the student reaches the end of the funding the viva is an additional mine-field as the examiner is likely to be another specialist.  The examiners specialism will enable them to either down-grade the students work in their own field, or not enable them to fully understand the major field of the student if it’s not their own.

In my experience this situation results in a very stressful environment throughout the Ph.D, with a continual lack of support in all of the component fields. There is a feeling of not developing a Ph.D level expertise in any particular area, and not contributing anything worthwile to the scientific community.  A lack of publications due to inherant technical problems with a poorly conceived inter-disciplinary research project then hampers any feeling of career development or planning for the future.

I think programs such as Bio-X at Stanford, detailed in the Nature paper are much better as the PI has grant funded inter-disciplinary research and the student then aquires the particular skills as and when required through collaboration between labs.  The inter-disciplinary training then comes from the nature of the host lab and feeds into the overall research objectives of the PI providing publication opportunities and post doc support.  It is not as some vague concept that orbits around the student.

An observation that seems to clash with the philosophy of the training center is that inter-disciplinary projects that are beginning to appear are employing specialists to contribute as part of a team.  Systems biology projects are recruiting modellers for the computer science component, and collaborating with dedicated wet labs for their parameter data.  The wet labs are then employing dedicated mass spectrometrists or microscopists to gather that data.  I have yet to see a wet lab systems biology project requiring biologists with modelling capability, or vice versa for the dry labs.  They are not recruiting for mixed skill-set researchers.  I believe the skills training provided by doctoral training centres is nice to have but is not required, and will actually hinder future job prospects for graduates.

I have yet to see a lab looking for a jack of all trades as the PI knows they are master of none.

ELRIG Drug Discovery 2011

ELRIG’s Drug Discovery 2011 conference is approaching (7th – 8th September) at Manchester Central. The annual Drug Discovery conference is a massive event with dozens of exhibitors from the lab automation industry. If you haven’t been to one it’s well worth attending, just to see the myriad of gadgets on display. There are also a large number of talks on academic and industrial applicatons of high throughput automation in research. The event is free to attend so there’s no reason not to skive off from work for a couple of days!

There’s more information available over at www.elrig.org.