Project Launch. 1. Which Other Epichem Compounds to Screen?

mattodd commented 6 years ago

As described in the preprint of the paper at project launch, there exists a large library of relevant analogs at Epichem, in Perth (Western Australia). A subset (about 5%) of these compounds were evaluated for the paper. Based on the data available (in the paper, in the Master sheet and shown graphically below), the question is: Of the remaining compounds in the Epichem library (contained in a sheet in the Master List, which compounds (if any) should be evaluated next?

Ideal situation: we identify 20 compounds with promising characteristics (potentially active based on data obtained so far, good pharmacokinetic properties) and these are shipped to Wendy van de Sande's lab in Rotterdam. Notice that in the paper, it appears we need to keep the logD low, to promote in vivo efficacy. This should be a design criterion.

structures of all fenarimols copy 2

Chemdraw of Epichem compounds evaluated: Structures of all Fenarimols copy.zip

MFernflower commented 6 years ago

Thoughts everyone? (please excuse the use of MSPaint)

Update 2/20/18: compounds that closely resemble the top most two have already been tested and shown to be non-potent - I apologize for the error @mattodd 36058673-de527e6e-0df4-11e8-835d-dc2d3af596ce

I would also recommend testing BS0446

MFernflower commented 6 years ago

The cyclopropylamide EPL-BS0549 would be a good screening candidate IMHO @mattodd

bendndi commented 5 years ago

Hi all, Here are the results from the latest batch of Fenarimol analogues to be tested.

new fenarimol analogue results nov2018

And here is the file with the data New data 1st Nov.xlsx

bendndi commented 5 years ago

From what I understand from Wendy, we may be able to do a larger number of Epichm compounds through the 25 and 100 uM single point assay - maybe as many as 96. How should we choose these?

mattodd commented 5 years ago

Great. I guess we need to identify the most similar molecules to those giving best potencies, while keeping firmly in mind the physicochemical characteristics we need for in vivo potency? Good to hear there's scope for Round 3 testing of Epichem library.

bendndi commented 5 years ago

@wwjvdsande confirmed we can probably aim for 90 compounds from the Epichem list

MFernflower commented 5 years ago

@bendndi are those observed results from assay of those compounds vs target fungus in vitro? @mattodd

MFernflower commented 5 years ago

EPL-BS0439 looks promising as it is only differs by one atom to a proposed compound I thought up in march ("Mandrake Urea 2") @mattodd @bendndi

drc007 commented 5 years ago

Great. I guess we need to identify the most similar molecules to those giving best potencies, while keeping firmly in mind the physicochemical characteristics we need for in vivo potency? Good to hear there's scope for Round 3 testing of Epichem library.

If I can play Devil's Advocate, if we only select the most similar molecules we actually learn very little extra knowledge. If we explore molecules that have points of divergence from the existing leads we may learn something new.

wwjvdsande commented 5 years ago

I agree, if we want to define the chemical properties needed to penetrate the grain and have in vivo efficacy we need a diverse sub-set of compounds. They should be soluble in our assay to, meaning a maximum of 2% DMSO is allowed.

bendndi commented 5 years ago

I'm also in agreement with @drc007 and @wwjvdsande - there is a big advantage to spreading the net wide within the space. Since @wwjvdsande has proposed we could test up to 90 compounds in the next iteration (at least at 25 um and 100uM), I'd suggest we go with both proposals - a tight set of compounds as propsoed by @mattodd and a broader set.

mattodd commented 5 years ago

A good plan. Agreed. We can obviously use this to test the predicted logD requirement for in vivo potency (described in the paper, Figure 10).

Just as a reminder on resources, the Master List of mycetoma molecules is here. There are sheets for all molecules evaluated as part of MycetOS and another sheet for the Epichem library. So: up to 90 diverse compounds, as well as a subset of those around the most active. We have proposed that in vivo activity is improved by lower logD, i.e. 2 or lower.

MFernflower commented 5 years ago

Interesting how less oily compounds seem to be more potent considering the fungal grain is lipid rich @mattodd

drc007 commented 5 years ago

@mattodd @wwjvdsande Since no one seems to have made a selection here is mine. I combined all the data on the google sheets and then did a K-means clustering to generate 80 clusters.

As you can see there are a few big clusters but only the highlighted have known actives (those which have quoted IC50). I then selected the centroid of each cluster, then selected a couple extra from the clusters that contained known actives.

The following list gives the identifiers from the table and the cluster number.

Other_ID,COMPOUND_ID,CLUSTER EPL-BS0118,SCYX0001294209,71 EPL-BS0282,SCYX0001402170,70 EPL-BS0516,SCYX0001454117,39 EPL-BS1025,SCYX0001697549,11 EPL-BS1365,SCYX0001810521,64 EPL-BS0010,SCYX0001249371,58 EPL-BS1483,SCYX0001808298,11 EPL-BS1482,SCYX0001808298,11 EPL-BS1477,SCYX0001800292,29 EPL-BS1480,SCYX0001800294,11 EPL-BS1105,SCYX0001729589,79 EPL-BS1107,SCYX0001729591,66 EPL-BS1440,SCYX0001845811,62 EPL-BS1439,SCYX0001845810,59 EPL-BS1080,SCYX0001724344,32 EPL-BS1108,SCYX0001511705,65 EPL-BS1090,SCYX0001724943,72 EPL-BS1089,SCYX0001724942,77 EPL-BS1091,SCYX0001724944,61 EPL-BS1072,SCYX0001720698,33 EPL-BS1044,SCYX0001713040,31 EPL-BS1008,SCYX0001683146,25 EPL-BS0863,SCYX0001602412,4 EPL-BS0922,SCYX0001610441,50 EPL-BS0930,SCYX0001613518,36 EPL-BS0928,SCYX0001613516,10 EPL-BS0839,SCYX0001596472,28 EPL-BS0881,SCYX0001607268,2 EPL-BS0874,SCYX0001607265,26 EPL-BS0810,SCYX0001576381,49 EPL-BS0799,SCYX0001574846,27 EPL-BS0761,SCYX0001555054,44 EPL-BS751,SCYX0001551252,57 EPL-BS0756,SCYX0001551254,7 EPL-BS0747,SCYX0001551242,48 EPL-BS0738,SCYX0001551240,7 EPL-BS0575,SCYX0001464842,1 EPL-BS0603,SCYX0001465511,5 EPL-BS0642,SCYX0001489898,40 EPL-BS0613,SCYX0001467766,9 EPL-BS0587,SCYX0001465079,7 EPL-BS0560,SCYX0001462989,60 EPL-BS0561,SCYX0001462990,38 EPL-BS0558,SCYX0001460910,12 EPL-BS0535,SCYX0001459256,34 EPL-BS0528,SCYX0001459242,24 EPL-BS0523,SCYX0001454263,6 EPL-BS0499,SCYX0001449903,22 EPL-BS0493,SCYX0001449878,7 EPL-BS0476,SCYX0001449766,17 EPL-BS0449,SCYX0001435848,15 EPL-BS0431,SCYX0001435830,13 EPL-BS0388,SCYX0001434961,8 EPL-BS0362,SCYX0001434935,19 EPL-BS0361,SCYX0001434934,46 EPL-BS0356,SCYX0001434929,23 EPL-BS0358,SCYX0001434931,20 EPL-BS0348,SCYX0001434921,56 EPL-BS333,SCYX0001433766,51 EPL-BS0193,SCYX0001335908,14 EPL-BS0317,SCYX0001423067,18 EPL-BS0310,SCYX0001423060,37 EPL-BS0207,SCYX0001336031,16 EPL-BS0277,SCYX0001402165,53 EPL-BS0269,SCYX0001392069,42 EPL-BS0271,SCYX0001392070,0 EPL-BS0255,SCYX0001388155,0 EPL-BS 254,SCYX0001388154,0 EPL-BS0240,SCYX0001388140,3 EPL-BS 158,SCYX0001322027,9 EPL-BS0208,SCYX0001336032,3 EPL-BS0200,SCYX0001335915,52 EPL-BS0196,SCYX0001335911,78 EPL-BS0188,SCYX0001332844,0 EPL-BS0170,SCYX0001325592,45 EPL-BS0160,SCYX0001322031,63 EPL-BS0150,SCYX0001317218,54 EPL-BS0115,SCYX0001294206,21 EPL-BS0132,SCYX0001303288,43 EPL-BS0114,SCYX0001294205,35 EPL-BS0107,SCYX0001278550,47 EPL-BS0098,SCYX0001268517,55 EPL-BS0087,SCYX0001268115,69 EPL-BS0088,SCYX0001268116,41 EPL-BS0090,SCYX0001268118,30 EPL-BS0080,SCYX0001267127,0 EPL-BS0038,SCYX0001255545,3 EPL-BS0052,SCYX0001255615,67 EPL-BS0018,SCYX0001246799,68 EPL-BS0014,SCYX0001246791,75 EPL-BS0013,SCYX0001246788,76 EPL-BS0012,SCYX0001246786,74 EPL-BS0005,SCYX0001246776,73 EPL-BS1387,SCYX0001813380,11 EPL-BS1392,SCYX0001813385,11

mattodd commented 5 years ago

Very nice @drc007 . So you mean to say that your selection above is representative, thereby meeting the criterion that we try compounds that are both similar and not, alongside a few extras that are similar to compounds with known IC50s, to check. If so, that seems like a solid way forward. @bendndi ?

drc007 commented 5 years ago

That is correct

wwjvdsande commented 5 years ago

Great @drc007 . Thanks for generating a list. I'm not a chemist so I leave the debate on the selection of the compounds based on their chemical properties to those who do. We would be able to screen the compounds against Madurella mycetomatis after the Christmas break.

bendndi commented 5 years ago

@drc007 I like the way you've done this and am happy to go with this.

If however we want to have some further consensus / input: I was in the middle of making a selection of my own but was looking at covering diversity across physchem properties rather than structural clustering. I was also wondering about limiting a larger subset of the selection purely to the compounds with piperidine type motif. How about I complete my selection, then we can compare the two approaches and either select one approach or take some from both (i.e. could you reduce your selection easily to 45 examples, I'll pick 45 examples using my approach, and we combine them?

wwjvdsande commented 5 years ago

@drc007, @bendndi. I think it is a good suggestion to make two independent lists and compare them.

drc007 commented 5 years ago

@bendndi @mattodd @wwjvdsande It will be interesting to see if there is any overlap.

wwjvdsande commented 5 years ago

@mattodd @bendndi @drc007 Has anybody checked for the overlap. It would be good to take this project further and screen additional compounds. I would be able to screen them in March

bendndi commented 5 years ago

@mattodd @wwjvdsande @drc007 I've run my analysis and selected 90 compounds. My process was as follows: I split out all the compounds which had the same particular core of interest, namely N-CH(Ar1)(Ar2). From this set of around 380 compounds I ran a structure based clustering with quite high similarity threshold and selected all the singletons and all the cluster centres. This gave me 58 compounds. I then went back to the full data set and removed all the compounds we've already tested. I then clustered based on properties only (calculated LogD, LogP, TPSA, flexibility, rot bonds...). I took the cluster centres and added these to the structure cluster centres. I then added in about 10 of the most diverse low LogD compounds. This tool me up to 90 molecules.

Overlap with Chris' set is minimal, I think about 10%

bendndi commented 5 years ago

@wwjvdsande I would suggest we go ahead an check @drc007 list first. We can then follow up with my complimentary list at a a later date. I'll ask Epichem to send these on to you.

wwjvdsande commented 5 years ago

@bendndi @drc007 that's OK by me. Keep you updated when we receive the compounds

wwjvdsande commented 5 years ago

@bendndi @drc007 @mattodd and all those who are following this project. We received the epichem compounds and tested them in 100 uM and 25 uM concentrations against Madurella mycetomatis. Out of the 61 tested, 8 were inhibited at a concentration of 25uM. These were: EPL-BS38 EPL-BS114 EPL-BS115 EPL-BS118 EPL-BS132 EPL-BS240 EPL-BS271 EPL-BS1025 Compounds which only inibited at 100 uM and not at 25 uM were EPL-BS269, EPL-BS493 (also showed some inhibition at 25uM but just above our threshold), EPL-613, EPL-761, EPL-BS839, EPL-BS922 and EPL-BS751. For the compounds inhibiting at 25 uM we will determine the IC50 in the upcomming weeks. MycetOS master excel updated.xlsx

MFernflower commented 5 years ago

@wwjvdsande Fungicidal or fungistatic?

wwjvdsande commented 5 years ago

For now we only tested inhibition. We did not check if it was cidal or static. When we determine the IC50 we will determine that as well. My guess would be static as the target is similar to that of the azoles and those are static. However, it will be good to determine it anyway. So we will include that this week.

MFernflower commented 5 years ago

Quick and dirty image depicting the latest structures shown to be effective vs the fungus (most of this new batch are quite similar to the Fenarimol parent)

MFernflower commented 5 years ago

@wwjvdsande would you be able to get your hands on some paclobutrazol? It's pretty much the flexible frame analogue of fenarimol

http://www.chemspider.com/Chemical-Structure.536024.html

MFernflower commented 5 years ago

The diaryl tertiary alcohols interest me the most as they can be cranked out en masse @mattodd

diaryltert

drc007 commented 5 years ago

@wwjvdsande Great to see that there are a number of actives, looking forward to seeing the titrated results.

mattodd commented 5 years ago

Wonderful @wwjvdsande will try to digest the new data. Yes, you can drag excel files onto comments. I've also uploaded the sheet to the Biological Evaluations folder (you can do this too):

https://github.com/OpenSourceMycetoma/Series-1-Fenarimols/tree/master/Biological%20Evaluation/2019%20Mar%20Wendy%20Epichem%20Data

but if you could add the data to the primary sheet we maintain of all MycetOS data, that will help a great deal. Ideally in the main sheet of MycetOS molecules, rather than as a separate tab?

wwjvdsande commented 5 years ago

We now also determined the IC50, IC90 and MIC for those compounds which displayed activity. We determined the MICs also for two additional strains.

MycetOS master excel updated.xlsx

MFernflower commented 5 years ago

@mattodd @drc007 Can somebody kindly render this data as a graphic? I'm currently on holiday

bendndi commented 5 years ago

Hi all, I''ve updated the master list with most recent results.

https://docs.google.com/spreadsheets/d/1YhK-3i2KwuVabo1GbZSgVjAUbavEICCMKi5v-EhNq80/edit#gid=954203120

I have a couple of comments / thoughts / questions:

@mattodd The following compounds seem to have been made but not tested, is that correct?:

HPD17_2F3 HPD20_1 HPD25_1 HPD26_1 HPD27_2 HPD28_1F2 HPD28_1F3 HPD29_1 HPD30_1

@wwjvdsande / everyone: Can we ensure that when communicating data / reporting we use exactly the ID numbers that are used in the sheets. Somehow around 70% of the Epichem ID's in this latest round had been altered somewhere between the actioning and the reporting, so for example "EPL-BS0038" became "EPL-BS38". This makes it extremely cumbersome and perilous to correlate data between batches and can lead to major errors i.e. assigning incorrect results to structures. I'd say this is essential to smooth running of the project (I managed to revert and reapply the correct IDs in the data sheet, but it took a long time and I cannot guarantee with 1005 certainty that an error hasnt slipped in somewhere). I'd say I'm 99.9% sure its ok.

also, if possible when reporting data can we report the SMILES string alongside the compound ID. I appreciate this may look like gibberish string to the non chemists, but to us chemists it's the golden nectar of science and including it in the reports makes our lives so, so much easier (and eliminates errors / brings a good check/balance to the reporting) :-)

Also, the master list (google doc) does not have the level of granularity that is in Wendy's excel master lost. Main thing is Wendy is reporting n=3 on the DRCs, and then a median IC50 / IC90, wheras the google docs master list seems to only have space for 1 example of a DRC. When I updated I just used the n=1 data from Wendy's file, since data seems to be quite tight. We may want to consider how we handle this going forward.

wwjvdsande commented 5 years ago

@bendndi: I'm sorry for this. I copy pasted the names from the excelsheet provided by Epichem when they sent the samples just to avoid getting mixed up numbers. In that list EPL-BS38 was there. The smiles were not present in the list of Epichem. Main problem is that we don't have funding for this project of course so everyone has to do these things in between. If I would have a person dedicated to this work it would indeed run things more smoothly.

bendndi commented 5 years ago

@wwjvdsande fully understand! I didn't intend this to come across as laying blame with anyone, I just wanted to highlight the importance that within the community we try to avoid these issues. I had kind of guessed that the issue lay with the lists sent by Epichem as you have pointed out so if anything the blame is on our side for not having our batch IDs aligned with those at Epichem...

In future I'll also make sure Epichem / other suppliers include SMILES strings in their shipment lists.

mattodd commented 5 years ago

@bendndi Hi - your question about those compounds that appear to have been made, but not tested. These were the ones (shown here) that Hung (@fantasy121 ) made after his thesis was submitted. They were shipped to @wwjvdsande and, I think, evaluated. Wendy - you may have sent me the results by email. I'll check, in case I didn't upload them. If you could also check, that'd be great. Are these already in the Master List? They're not in your Master List, Wendy, in the Sydney tab.

For the most recent new data, above, @bendndi did you say all the new data are in the online Master List? We shipped 90 compounds, I think (?), but @wwjvdsande 's Excel sheet has 108 compounds, and I'm not clear which entries are the new ones...

I'm asking for clarity on this because @dmitrij176 is about to generate the graphical image of the actives and inactives, so that we can all plan next steps, and I want to be 100% sure we're all looking at the same dataset.

bendndi commented 5 years ago

@mattodd I confirm that all new data is available in the updated master list google doc, with the exception of that set of compounds from Hung.

The same data set is also available as an sd file here

wwjvdsande commented 5 years ago

@mattodd @bendndi we indeed screened Hung's (@fantasy121) compounds and I mailed the MICs to Matt. I will add the raw data to the table. Hopefully i will be able to do that today, otherwise it will be on thursday. I will also highlight the newest compounds in the sheet.

wwjvdsande commented 5 years ago

@mattodd @bendndi I added the results we had on Hung's (@fantasy121) compounds. I could not find the 100 uM results in the file. I guess we immediately went for MICs. To highlight which were the newest EPICHEM compounds I added new under comments. I added it in the google doc.

dmitrij176 commented 5 years ago

Good day @wwjvdsande. I have a question regarding new data. During analysis, I came across a number of compounds that appear to have negative values (eg: EPL-BS0038). I have highlighted these in blue in the Google doc. Could you please be so kind to clarify whether these structures are active/inactive or the negative value has any other meaning? At the moment, I am not entirely sure which group to assign them to.

wwjvdsande commented 5 years ago

@dmitrij176 the negative values mean that they are inhibited. The negative value means that the optical density obtained with the viability dye was lower as that of medium with the viability dye. This results in a negative growth percentage. Every value below 20% (even if it is negative) is considered as (complete) growth inhibition in this assay. Therefore the negative values can be considered as no growth and the compounds as active.

wwjvdsande commented 5 years ago

If negative values don't work in your calculations you can also consider them as zero. Biologically seen it does not make a difference as both are considered no growth.

dmitrij176 commented 5 years ago

Thank you @wwjvdsande

drc007 commented 5 years ago

@mattodd @wwjvdsande @dmitrij176 @bendndi Here is an image showing all the results with "New" in the notes column. I've also included activity at 100uM and 25uM. Might be worth titrating EPL-BS1025 and EPL-BS0118.

wwjvdsande commented 5 years ago

@mattodd @wwjvdsande @dmitrij176 @bendndi good suggestion. We will test the hits in more details. EPL-BS1025 was also tested in the original paper and was already tested in the Galleria mellonella larvae. It enhanced the survival of infected larvae significantly.

wwjvdsande commented 5 years ago

EPL-BS1025 is twice in the list now as we tested two different batches of it.

drc007 commented 5 years ago

I noticed, but I'm a belt and braces type of person. Couple of repeats using different samples gives more confidence.

mattodd commented 5 years ago

Hi @bendndi @drc007 did you notice the data from the other compounds @wwjvdsande previously evaluated, those made by @fantasy121 here. Rows 46-54. The data show that HPD20_1 is quite potent. That's the analog with a MeO in place of the OH on the central carbon. Interesting, no? Could be a prodrug for the alcohol, of course, depending on how good a leaving group that central OH or OMe is.

OpenSourceMycetoma / Series-1-Fenarimols

Project Launch. 1. Which Other Epichem Compounds to Screen? #1