The good start I was hoping for on Monday morning didn’t quite come into fruition. But, it wasn’t a total disaster. The job hadn’t failed or been killed, it just hadn’t finished running yet. Which was fine… except I wasn’t sure exactly how long things were allowed to run until they were killed by the job scheduler. I’d heard ~2 and a half days, and as the clock approached almost 3 days of run time I started to get very nervous, especially since the log files were indicating that the jobs were only about 65% complete (turns out it took 5 days but it was allowed to continue running so I’m taking that as a win).
After working on running some other software, I had a meeting with the production bioinformatics team in the afternoon to find out a little bit more about what their pipeline involves. This is the team that processes the samples that have been sent to the sequencing service and are primarily whole-genome diagnostic samples. The extent of the bioinformatics they do is quality checks, variant calling and some annotation. They also filter the variants using some common filters so the interpretation team only receive a list of hundreds, rather than hundreds of thousands. They also convert the vcfs to a Gemini database to aid the interpretation team with querying the variants found in each sample and have created some graphical user interfaces to further facilitate downstream analysis. It has been stressed to me that the Garvan is a research institute, not a diagnostic service- despite having some clinical aspects, which explains why there is limited filtering and prioritisation of variants in their standard production pipeline. But it was interesting to compare it to our diagnostic pipeline in Exeter, where a huge part of what we do is incorporating annotations and rule sets to create the smallest feasible list for our genomic scientists to analyse.
On Tuesday I had a tour of the sequencing labs. It was interesting to hear about all the projects they’ve got going on and see their sequencing facilities. At the moment they have far more sequencing power than they actually require and it will be interesting to keep up with how they move forward as their sequencing machines approach the end of their life spans. They’ve automated a lot of their lab work using robots which is cool, it means they have a lot going on without the need for a huge team and it was great to see some of the things they’re proud of, were things that are also done in the lab in Exeter.
On Wednesday I had a catch up with a pathologist called Leslie. I say pathologist, but he has expertise in bioinformatics, interpretation and the clinical side of things too. He’s developed a machine-learning algorithm to help narrow down the variants they get at the end of their bioinformatics pipeline to decrease the amount of manual interpretation they need to do. He showed me some statistics that painted a pretty impressive picture for maintaining sensitivity while significantly decreasing the number of variants to analyse at the end. Apparently, this tool actually outperforms their pathologists when it comes to identifying the most likely causative variant. It’s based on two orthogonal pipelines- one based on genomic annotations and filtering and the other based purely on phenotype. The overlap in variants at the end of these two pipelines are the most likely to be causing the condition.
Thursday and Friday flew by, having decided that it was about time to start writing up about some of the work I’ve done, documenting commands and scripts so they can be used in the future and actually making sense of some of the data generated.
Despite having been documenting my experience as time has passed, it still seems unbelievable that I’ve already completed the final stretch of my elective. Everyone in the office seemingly felt the same as I received many a “so it’s your final week, hey?!” and “wow that’s gone quickly” throughout my last week.
On Monday morning, I had to give a presentation on the work I’d done with Andre over the previous 5 weeks. I was really nervous about giving this presentation, as per usual, but Andre presented with me on some of the bits that he had more involvement with. It was a really casual presentation- more of a discussion and I was able to answer the questions that got thrown at me so overall I was happy with how it went.
Fortunately, having already summarised our findings and presented them on Monday morning I had the rest of the time in the office to write up some reports and make a presentation for the lab back in Exeter. Not only did I have to write up a report for my elective competencies, but one of the conditions of the travel grant jointly awarded to me by ACGS and BSGM was to provide them with a report as well. So, I was grateful to have a bit of time at work to do this. However, this wasn’t the whole week as I had registered to attend a conference in Canberra from Wednesday-Friday.
The conference was the Australasian Leadership Computing Symposium and offered 5 streams for computing in different scientific areas- one of them was genomics. Thursday was the main day of genomics talks and it was very interesting to hear discussions around compute resources and the challenges posed to the genomics community by “big data”. In a research setting there was discussion around optimising pipelines to get them to run faster due to limited resources as well as the cost of “compute hours” draining their grant money. This is highly transferable to a clinical setting, where halving the compute time can save more than money. It can potentially save a life. Overall it was an exciting couple of days with an enjoyable networking dinner on Thursday evening where I had the opportunity to meet scientists in research and industry from many different backgrounds.
So I guess this is the part where I look back and tell you what I’ve learned from my elective and what I’ll take away from the experience. But I think a lot of the experience can’t be explained with tangible learning outcomes. Of course, I’ve learned a heap of new skills which I’m thrilled about and I’m sure they are going to be useful in my practice as a qualified scientist. But, being in such a different environment has enabled me to shift my thought process when addressing new problems and be more open to seeking help and discussion from the community and people I don’t know that well, if at all. Through discussions and the conference, I’ve also got a lot of food for thought on possibilities for improving the services we offer now and bridging the knowledge gap between bioinformaticians and the scientists who receive the data we’ve processed. I am so unbelievably grateful to the team at the Garvan and KCCG for hosting me and being so open to discussions and my questioning, particularly Andre who really made sure I got the most out of my experience. Also to ACGS and BSGM for the travel grant which really helped towards relieving some of the stress that comes with organising an overseas elective. Also to my department for supporting me and allowing me to take the time to do this.
My final thought on the elective is this: you don’t have to go somewhere exotic or do something cutting edge to have a rewarding elective. Wherever you go and whatever you do, you get out what you put in. The first couple of weeks can be tough adjusting, but stick it out because it will absolutely be worth it.
I hope you’ve enjoyed reading this mini-series as much as I enjoyed writing it. I’m now off on annual leave (you know I couldn’t go to Australia and not take some time off!) so it might be a while before you hear from me again! But thanks for reading and until next time!