tdido

Spanish National Cancer Research Centre (CNIO) | Madrid, Spain | HYBRID | Systems Administrator (Infra, Storage and Backup) | Full-time, permanent | €45k-65k gross/year | Apply by Nov 11

CNIO is one of Europe's leading cancer research centres. We're hiring a systems administrator to run the infrastructure behind our research: petabyte-scale storage, backup and archive (including tape), Linux and Windows servers, and AD/M365 accounts. You'll work alongside senior colleagues, with room to take ownership of core systems as you grow.

What we're looking for:

- 3+ years of hands-on Linux and Windows server administration (not desktop support), including AD and file-share permissions

- Experience with enterprise storage or backup/restore

- Scripting (Bash, Python, PowerShell)

- Someone who cares that restores actually work

- Bonus: large-scale storage, tape and archiving, M365, virtualisation, config management, ENS or ISO 27001

- English and Spanish at professional working level

Apply by 11 November: https://www.cnio.es/empleo/systems-administrator-infrastruct...

Informal enquiries: tdidomenico-at-cnio-dot-es

Spanish National Cancer Research Centre (CNIO) | Madrid, Spain | HYBRID | Senior Scientific Computing Engineer | Full-time

CNIO is one of Europe's leading cancer research centres. We're looking for a senior scientific computing engineer to join the IT, Scientific Computing and AI area and take ownership of the infrastructure behind our research, from genomics pipelines to AI models. You'll be the senior scientific computing person in the area, reporting directly to me (the area head), and you'll help define where scientific computing at CNIO goes over the next few years.

What you'll do:

- Shape the roadmap for scientific computing at the centre and drive it forward

- Own and evolve our HPC cluster (Slurm): scheduling, monitoring, capacity planning, user support

- Build out our GPU / AI infrastructure as the centre's ML workloads grow

- Work with researchers to turn their analyses into robust, reproducible pipelines and tools

- Optional: co-author research papers

What we're looking for:

- Several years running Linux HPC environments in production (Slurm or similar)

- Comfortable with automation (Ansible, Python), containers (Apptainer/Docker) and workflow managers (Nextflow, Snakemake or similar)

- Someone who wants to set direction, not just keep the lights on

- Bonus: Ceph, Kubernetes, GPU stacks, a bioinformatics background

- Spanish not required

If interested: tdidomenico-at-cnio-dot-es.

You can go a very long way just mapping your DNA reads to a reference genome [0]. This reference will have associated annotations of different types, e.g. mutations, which automatically give you relevant information about your own genome by being present in it (or not).

Mapping is a fairly straightforward process where SOTA software is FOSS (as is most bioinformatics software). Accessing databases can be as simple as linking your data to the UCSC genome browser [1]. Of course, if you want to go the manual way, the sky is the limit.

[0] https://www.gencodegenes.org/human/

[1] https://genome.ucsc.edu/

See here for updates: https://bsky.app/profile/matrix.org/post/3lxuslbzjuc2t

That's actually pretty much what Andrej Karpathy mentions as a mitigation for hallucinations here:

https://m.youtube.com/watch?v=7xTGNNLPyMI&t=5400s

Spanish National Cancer Research Centre (CNIO) | Madrid, Spain | HYBRID (2d/week on-site) | HPC Engineer | Full-time

We're looking for an HPC engineer to join us at the Bioinformatics Unit to help us boost the scientific computing infrastructure at our research centre.

What you'll do:

• Administer & grow our HPC cluster (Slurm)

• Help implement our new storage system (Ceph)

• Optional: write tools and/or co-author research papers

Apply: https://www.cnio.es/empleo/advanced-degree-for-the-bioinform...

Questions: tdidomenico@

Having a 3D printer available, I've been using these two projects to set up my modest workshop and it's worked out great.

https://github.com/jeffbarr/gridfinity-catalog

https://github.com/aderusha/DDD-Printable-Wall-Control-Syste...

Very much so. As a bioinformatician, for me conda has become a central part of my workflow as a generic package manager through the conda-forge and bioconda channels. I know a few university HPC clusters that use it to handle software instead of e.g. environment modules.

I sync my journal directory using syncthing and it works reasonably well.

This is used in both bwa [1] and bowtie [2], two of the most popular DNA sequence aligners.

[1] https://github.com/lh3/bwa

[2] https://github.com/BenLangmead/bowtie

I would easily expect even worse than that any day during peak hour in any of the larger stations here in Madrid.

It won't be perfect, but you will have a fair amount of places where it can and will be implemented. Our research centre has central A/C and they announced the change a few days ago.

Both more powerful and less glamorous than Mare Nostrum 4, which is located inside a former chapel: https://www.silicon.es/wp-content/uploads/2017/06/2017_BSC_S...

You can use "-i" for in-place password generation, which will just replace the first line.

El Diablo and Murcielago were also fighting bulls.

Not paid well at all, so compensation can't be a driver. That doesn't mean you can't find decent wages, but the average is low.

If you're still interested feel free to email me and I'll be happy give you some pointers. Email in my profile.

Certainly not because of compensantion, but I switched to academia after 10 years as a consultant by doing a PhD in bioinformatics. I now work in a cancer research centre.

There is a serious lack of people with computational background in the field, very hard to compete with the market in terms of compensation. So we are left with the rare cases where the motivation of the type of work beats that, or the few places (e.g. Broad in the US) that actually have special consideration in terms of salary for computational biologists.

Going through a PhD is probably the one thing that would scare people away as compensation is usually dowright ridiculous, but as someone else mentioned you can find plenty of positions that will but require it. Pretty much all of the students from our Master's degree who don't even have a PhD find jobs in the sector.

I use https://www.mkdocs.org/, which is markdown-based. It's simple to get started and customizable enough if needed.

Seems like there's a conda package in the works: https://github.com/bioconda/bioconda-recipes/pull/29660

See also:

https://dl.acm.org/doi/pdf/10.1145/3360551

https://www.nature.com/articles/s41587-021-00985-6 (paywalled)

Actually, for the use case of genome assemblies you can compensate for the error rate with depth of coverage. Long reads are already the state-of-the-art for assembling genomes and are allowing us to get information that is simply invisible using short reads.

https://www.biorxiv.org/content/10.1101/2021.05.26.445798v1

They seem to be under the FCA: https://transferwise.com/help/articles/2932699/what-is-the-f...

Do you mean they don't subscribe to the 80k protection or something like that?

Ah, sorry about that. I'm not getting a paywall for whatever reason.

We're certainly already there with food related companies: https://www.businessinsider.com/10-companies-control-the-foo...

Actually, one of the main features of this tech apart from the obvious size-factor is that it's a streaming process. You can analyse data on the fly and decide when to stop the run. Wash the flowcell, and use it for another sample. Eventually the pores die, yes, how fast depends on the sample type. I think they guarantee 48 hours or something of the sort.

The expensive part is not the chemistry. Each flowcell has a very expensive piece of metal that senses the very small current variations that each kmer causes when going through each pore. They've actually come up with a device (horribly named "flongle") that has the same shape of a flowcell but no pores, and the mini flowcell it uses is ~90USD (against ~900USD for a full flowcell). Of course, yield is much lower.

Nope. You're working with kmers. I think it's 6mers in the current models. It's good because you get redundancy as you move, but coupled with the fact that you can't control dwelling time it makes repetition hard to handle.

Still, some of the errors can be compensated for with more coverage. So if you can manage 20-30X you're left with the homopolymer problem (nanopores can't tell how long a stretch of the same repeated nucleotide is, because you can't control how long the sensed kmer stays in the pore), but lots of other types can be improved quite a lot.