Hi @vonboss, @ShayReza, @LucaChiesa, @jmchap, One question, prompted by what the NTC supplement showed. All 7,703,070 rows of PGK2_selection.parquet have count_PGK2 >= 1 - the minimum is 1, and there are no count_PGK2 = 0 rows. So count_NTC is reported only for compounds detected in the PGK2 arm. Summed over the main file, it is 5,649 reads; adding PGK2_NTC_supplement.parquet brings it to 248,922. The censored subtotal understated that arm by a factor of 44. count_PGK2_with_inhibitor sits in the same file under the same condition, so I assume it is censored the same way, which would make its 49,707-read total a subtotal rather than a depth. 1. Is there a corresponding supplement for the inhibitor arm - compounds with count_PGK2_with_inhibitor >= 1 and count_PGK2 = 0? @vonboss mentioned on 14 Aug having obtained that data from Baylor. Even "not before 1 Oct" would help us plan. 2. If not, could you confirm whether the released inhibitor column is censored to PGK2-positive compounds? That alone changes how the Hit Selection criterion should be read, and it is a one-line answer. For context on why this matters rather than being idle curiosity: fitting the observed inhibitor-to-target read ratio per target-count band on the released file gives 0.0034 at count_PGK2 = 1, rising to 0.38 at counts 101-1000. Since count_PGK2 = 1 covers 98.24% of rows, the recommended count_PGK2_with_inhibitor < 0.1 x count_PGK2 criterion is satisfied for most compounds whether or not competition occurred, while for the few hundred with high target counts it discriminates well. If the column is censored, the low-count end of that curve is partly an artefact. Thanks very much,

Created by Hsuan-Kai Wang DoingWell
Hi Organizers ( @vonboss, @ShayReza, @LucaChiesa, @jmchap), I’m following up on the earlier discussions about compounds absent from PGK2_selection.parquet and the control-arm supplements. While reconstructing the qDOS24 response pattern, I found two complete 68-compound families that are present in the enumerated qDOS24 library but have no rows in the released PGK2 selection file: - qDOS24-19-1-241 through qDOS24-19-68-241 - qDOS24-53-1-241 through qDOS24-53-68-241 These are 136 distinct compounds in total. I don’t want to assume that absence from the released file necessarily means a measured response of zero. Could you clarify: 1. Were these compounds physically present in the synthesized library and exposed in the PGK2 selection, or does their presence in the enumeration represent design space only? 2. If they were included, does Baylor or the challenge team retain any per-compound data for any subset of them—such as PGK2, PGK2-with-inhibitor, NTC, normalized Z-scores, or later assay results? Even a partial list of compound IDs and available measurements would be useful. 3. If no archived measurements exist, is there a realistic route for obtaining responses for this exact set in a follow-up experiment or individual assay? If so, who would be the appropriate contact, and what protocol would make the results comparable with the released data? I’ve read the existing threads on control-arm-only reads, inhibitor censoring, and absent sub-libraries; unless I’ve missed something, none resolves this narrower question. I can provide the complete list of 136 compound IDs if useful. Thanks very much, Julian
Hi @DoingWell, I'm glad you found the same thing as I did in the [earlier thread](https://www.synapse.org/Synapse:syn75349604/discussion/threadId=14426), always nice to have another pair of eyes confirm the same result. I'm honestly not sure if there is enough "with inhibitor" data to get reliable conclusions, even for the most enriched series. Best, Alex
@DoingWell 1) No, we are not able to provide the corresponding supplement for the inhibitor arm - compounds with countPGK2withinhibitor >= 1 and countPGK2 = 0? I originally thought that I had this data, but it was for an earlier replicate of this experiment and not comparable. 2) Yes, the released inhibitor column is censored to PGK2-positive compounds. I agree with you that this information would be useful sorry we were about to secure the data in time for the challenge.

Is count_PGK2_with_inhibitor row-censored the same way count_NTC is? page is loading…