I'm new in bioinformatics. I'm working in a clinical microbiology laboratory, and my chief wants to stablish a sequencing department for clinical and research purposes (whole genome sequencing for studying outbreaks, 16s metagenomics for studying microbiota, etc). We want to buy a desktop PC to do all this kind of bioinformatic analysis. For now, I'm using a VM in a laptop with W10, 8GB RAM, 512 GB SSD and Intel Core i7 7gen processor. I faced some memory issues in certain pipelines I run, and it´s time to move to a more powerful machine. Which specifications could you recommend me? I feel confident using a linux-VM in Windows, so I don't want to buy an apple/IOS machine for now.
I have read in other forums that It will be enough 16-32 GB of memory and 1-2 TB of storage...I'm scary about buying something that later doesn't fit our department needs.
Can you give some more details on the scale of what this department is going to be. Is it some smaller experiments from time to time or is this planned to be a facility that serves an entire hospital and external collaborators? Also, do you or anybody in the department has experience in handling and administrating large-scale sequencing projects and data volumes? Also, is there the hardware for secure backup of raw data?
Hi, sorry about not be as concrete as I should. Our goal is to support the needs of our clinical microbiology laboratory, therefore we would not have a constant activity on a large scale. None of us have broad experience in bioinformatics nor NGS (we about to finish the residency training program in clinical microbiology). Actually, our service doesn't have a DNA sequencer (we outsource the service). We only want to analyze raw sequencing data, in order to not to have to outsource the analysis too. We think it would be enough to buy one powerful computer to start...we can´t afford to build a workstation/cluster for now and we don't have enough informatic knowledge to do so.
I think you are getting a bit confused between IT requirement and bioinformatics here.
It would be better to speak to someone in the local IT team who would be presumably managing the system you are going to purchase. They would be able to advise you on the data handling and backup strategy and solutions. It would also help if you would provide them with details about approx size of the data being produced/analysed per year. It is a pure infrastructure/IT based questions and it is not possible for many of us to address such specific questions on this forum.
I'd suggest that you look at previous threads related to this question on the forum as it should give you an idea about the specification that you might want to go for within available budget.
It depends on what you want to do. In case you want to perform Whole Genome Sequencing metagenomics or metatranscriptomics, you would need quite a lot of RAM memory (specially for Kraken and CLARK), I would say more than 64Gb. It always depends on the pipeline used, megablastn + MEGAN is a slow approach but needs less memory (with 16-32Gb would be enough). Also, if you plan to sequence a lot, you might have a shortage of disk-memory with only 2Tb.
An alternative would be to use online tools to get the data analyzed, such as Gaia (Demo).
Getting your hands on as much RAM as you can will be the killer as others have said.
You will really want to look toward buying a "Workstation", rather than a desktop PC, since you will eventually find yourself needing closer to server grade components I suspect. A workstation is not necessarily the same as a server, so fulfils a happy medium I think.
Also, I'd strongly recommend just taking the plunge and getting a proper linux machine set up. VMs add overheads to your computation.
In terms of storage, I would also go quite big on this from the start. Hard drives are dirt cheap (not to mention a single even consumer grade hard drive is inching toward the 8-10Tb mark now).
It would probably be helpful if you could tell us your budget though.
I agree with the other. It depends on you plans. However, I see this whole genomic/metagenomics field is moving pretty fast forward. Today it is hard to tell what kind of analyses with what kind of software tools you are going to need in two years from now. Also, your sequencing strategy might change (short to long read shift) and then you need to adapt. If it is possible for you or your institution you could think about renting a computer in the cloud. It's quite inexpensive, you only pay when you need it, and if you need more cores or memory you simply seek more without the need to buy a new computer or extend your server and to setup your software tools again.
There are a lot of companies offering great services like AWS, IBM or Microsoft Azure. In my opinion, AWS is mostly ahead, and they have great big data analysis services, also for bacterial pathogen analysis. I was always a bit worried about data protection in the cloud, but specifically with AWS you can decide where you data is stored. They also have released a whitepaper about how you can do cloud computing in compliance with European law. So this is getting much better. Maybe worth for you to check it out.
Can you give some more details on the scale of what this department is going to be. Is it some smaller experiments from time to time or is this planned to be a facility that serves an entire hospital and external collaborators? Also, do you or anybody in the department has experience in handling and administrating large-scale sequencing projects and data volumes? Also, is there the hardware for secure backup of raw data?
Hi, sorry about not be as concrete as I should. Our goal is to support the needs of our clinical microbiology laboratory, therefore we would not have a constant activity on a large scale. None of us have broad experience in bioinformatics nor NGS (we about to finish the residency training program in clinical microbiology). Actually, our service doesn't have a DNA sequencer (we outsource the service). We only want to analyze raw sequencing data, in order to not to have to outsource the analysis too. We think it would be enough to buy one powerful computer to start...we can´t afford to build a workstation/cluster for now and we don't have enough informatic knowledge to do so.
I think you are getting a bit confused between IT requirement and bioinformatics here. It would be better to speak to someone in the local IT team who would be presumably managing the system you are going to purchase. They would be able to advise you on the data handling and backup strategy and solutions. It would also help if you would provide them with details about approx size of the data being produced/analysed per year. It is a pure infrastructure/IT based questions and it is not possible for many of us to address such specific questions on this forum.