Apple has released the new generation of Apple silicon MacBook Pros, and honestly I'm impressed. Do you think it's reasonable to purchase a MBP equipped with M1-max chip, 64G Ram, and 2T storage for single cell RNA seq data analysis?
I really wish any professional bioinformaticist from this forum can share their idea, or even post some testing videos when they get this new model.
Sometimes we purchase things because we like the idea of having them rather than considering their actual utility. In this case the purchase would fall somewhere in the middle.
It would be a fine overall computer that will give you years of service as long as you take care of it but don't buy it for specifically doing scRNAseq analysis. As others have noted, depending on the size of the datasets there may be parts of the workflow that you can certainly run on the laptop you describe. Once you have various count matrices those should be amenable to analysis with R on the laptop locally.
If you need a laptop at present and can afford to spend the extra money then you will not be disappointed.
Anandtech has a technical review of M1Max. It is a beast of a chip. A Mac Pro whenever it appears should have a chip better then this.
Note: Apple is not building these for bioinformatics. MBP's are being targeted to creative professionals (video/audio etc). Which is why M1Max gets double the GPU/encoding/decoding/memory, while its CPU core count stays the same.
It's way more cost effective to carry out heavy stuff on headless servers. They're also far more future-proof as things can be added/upgraded later on. With laptops you're generally stuck with the soldered components you initially chose.
I generally don't like to use laptops for serious bioinformatic work.
scRNAseq analyses that involve a large number of samples will require days or weeks of continuous heavy load on a single computer. A laptop isn't really designed for sustained heavy load. The CPU heats up and a laptop doesn't have good ventilation, and the CPU will throttle its own speed to cool itself down. The kind of sustained high temperature and heavy IO load can really shorten the laptop's lifespan.
I have another take on this: it's absolutely worth it. Running on HPC most often means non-interactive workflow where you run a script and can't see the result immediately (unless there's a high-mem Jupyter instance available). At the moment, I'm running Seurat with Jupyter on a virtual machine in the cloud. Integrating 30-40k cells requires about 25-30G of RAM. So with 64G you can process even more than that. Scanpy is much more modest, so you'll be able to analyse larger datasets.
Genomax made an in-depth review of apple M1: Apple M1 processor for bioinformactics . My impression is that it is a great all-around computer, but the heavy stuff should still be done on remote server.
Thank you all for sharing your opinions. I basically have calmed down, and in deed, i do agree that remote server can be more efficient and future proof.