My German cartesian mind (here we go with stereotypes) asks me to start with the first protein expressed by the viral genome of SARS-Cov-2, nsp1. That is, by far, not the low hanging fruit. But it is interesting to learn more about the roles of the different constituents of the virus. Nsp1 actually has also a few very interesting roles to facilitate the viral lifecycle in its host environment.

The bread and butter of every structure based drug design or drug ID effort is the need for 3D structural information of a target of interest related to the phenotypic outcome we want to alter. In the case of SARS-Cov-2 several experimental structures are already available in the RCSB PDB. Other a large scale fragment screens from the Diamond Light Source are slowly becoming available (available in the 3decision protease project for now) on public resources likes the RCSB PDB. Popular homology modelling services also already started to provide several high quality models for some of the yet not resolved structures of the viral proteome.

But let’s first have a look what the viral genome actually contains and what are currently pursued efforts.

Gabriella outlined in previous posts (part I & part II) a few things about SARS-Cov-2. Back at the time the crisis was still far away from our daily lives but things have changed dramatically during the last weeks. After our initial communication to make 3decision.discngine.cloud freely available for all Covid-19 related collaboration projects we were contacted independently by several volunteers from academic and industrial groups to collaborate on a global effort. At the mean time a lot of novel initiatives have seen the day and everybody is publishing blog posts (like I am right now), linkedIn articles, chemarxiv or bioarxiv preprints etc. It’s a little bit a scientific wild-west happening right now on a global scale.

At Discngine we are trying to provide the best services for our customers. In the pharmaceutical domain, we are led to help them to construct machine learning models for properties and biological activities prediction. Lately, we also tried various deep learning methods when it was possible...

As a follow-up to my previous article 2019-nCoV, structurally speaking, I thought I’d share some of the news on the protein structure side and some insights from the structural analyses we’ve done at Discngine.

I guess nobody has missed the fact that we are in the middle of a new Coronavirus outbreak. In the beginning, I followed the updates from far, without going in too much detail. That was before I saw a LinkedIn post on a homology model of a 2019-nCoV protease last week.

This time around you won’t see any source code. Sorry about that, next time maybe.

Here I’ll talk more about sometimes overlooked aspects when developing applications for R&D in life sciences. So I guess this post might be useful to our competitors or people developing applications within large corporations in the life science & pharma industry … anyway ;)

I’ll take the example of our current 3decision developments. You might have heard or read about 3decision already before. If not, check it out - 3decision .. it’s super cool!

I love MOE, it’s a great piece of software, but unfortunately it’s not free and now that I’m nor in academia, nor in a pharma company I find myself unable to further develop an approach I implemented in the past which is kind of frustrating to put it politely. The same goes for Amber.
So I started to look for alternatives, also under a gentle demand from Anthony (poking me regularly) who wanted to apply the whole approach in his fragment screening pipeline. That’s in the scope of the crazy impressive XChem project. If you haven’t heard / read about that check it out - it gives you a bit of an insight on what will be possible very near future in xray crystallography!

I think that’s my first RDKit post! So reason to celebrate!

Here' I’ll focus on a very nice feature available in Open Source Software and very useful in daily structure based but even ligand based drug design tasks.

The problem

How can I dock small molecules into a receptor and avoid as much as possible the “what is the right pose” problem?