PostEra

Filter out covalent ligands

COVID Test
#1

Hello,

First of all I really like the interactivity of the assay results/compound tracker. One thing that I’m missing is the distinction between covalent/non-covalent hits however. Would it be possible to filter on covalent compounds? Now you can add this as a property but I can not easily get a ranked list of the top non-covalent compounds in terms of pIC50?

Thanks!

#2

I combined the exported data from the activity and compound tracker pages to generate a ranked list with the additional covalent columns. The spreadsheet is by default ranked by pIC50 within the subgroups of covalent/non-covalent… is this something like what you were looking for Bart?

I’ll let @mc-robinson come back on adding the covalent tag directly into the activity data export.

activity_data_with_covalent_filter.csv (68.2 KB)

#3

Yes indeed, so being able to filter (in the top):
https://postera.ai/covid/submissions/compounds
on the Covalent_warhead column (like is now the case with assayed/ordered/made)

Thanks!

#4

Which is the difference between “covalent_warhead” and “covalent_fragment” columns? How are they estimated/calculated? I am also interested in filtering covalent inhibitors, so I assume that covalent inhibitors are those with “True” in the covalent_warhead column.

#5

Hi @santi.garcia-vallve, “covalent_warhead” indicates that the design itself contains a warhead. “covalent_fragment” is slightly outdated, but indicates that the design of a compound was inspired by a covalent fragment during the initial fragment expansion stage.

#6

Just to hi-jack this a little bit, would it alos be possible to add in to the activity data export the fragalysis code for those compounds with structural data? @mc-robinson

1 Like
#7

Thanks @Ben_DNDi, I’ll try to get that done and let you know when it’s available. For now, the list of all structures can be found in this file: https://github.com/postera-ai/COVID_moonshot_submissions/blob/master/structures/fragalysis_structures.csv

#8

Hi @mc-robinson, sorry for asking again about this subject. How could I calculate if a compound (not from the COVID Moonshot) contains such a covalent_warhead? Is it possible to know it from the smiles? Which rules do you use to estimate this field? Any help is appreciate. I am interested in knowing if the initial compounds of a VS contains a covalent_warhead.
Thanks

#9

Hi @santi.garcia-vallve, these were the warhead rules used, though please note that they are nowhere near exhaustive.

covalent_warhead_smarts = {
    'acrylamide' : '[C;H2:1]=[C;H1]C(N)=O',
    'acrylamide_adduct' : 'NC(C[C:1]S)=O',
    'chloroacetamide' : 'Cl[C;H2:1]C(N)=O',
    'chloroacetamide_adduct' : 'S[C:1]C(N)=O',
    'vinylsulfonamide' : 'NS(=O)([C;H1]=[C;H2:1])=O',
    'vinylsulfonamide_adduct' : 'NS(=O)(C[C:1]S)=O',
    'nitrile' : 'N#[C:1]-[*]',
    'nitrile_adduct' : 'C-S-[C:1](=N)',
    }
#10

Thanks @mc-robinson for the info. It will be very useful for us. I have detected that some of the filters (specially the nitrile and nitrile_adduct) has not passed to all the compounds. At the submissions.csv file, there are several compounds with some of these filters in blank (without the TRUE or FALSE value), so some compounds marked as covalent_warhead FALSE must be covalent_warhead TRUE. For example the compounds TRY-UNI-2eddb1ff-6, LOR-NOR-30067bb9-19, BAR-COM-0f94fc3d-7 and BEN-DND-61647d40-8 have a nitrile warhead not showed at the submissions.csv file