2018-05-22 Project Team Meeting - Agenda & Minutes
- Richard Norman (Unlicensed)
- Abhinandan Raghavan (Deactivated)
Date
Attendees
- Laura hook (Deactivated)
- Bryan Jones (Deactivated)
- Chris Lloyd (Deactivated)
- Carmen Nitsche (Deactivated)
- Abhinandan Raghavan (Deactivated)
- Kieran Todd (Deactivated)
Discussion items
| Time | Item | Who | Output |
|---|---|---|---|
| 17:00 - 18:00 (60 mins) | >'Basecase' scenario. >Upcoming comms, presentations. | All | >Draft ‘basecase’ scenario. List of assays in scope. >Outline and items we want input on.
|
Minutes
GSK feedback from discussions with legal and internal stakeholders
>Expectations: interested in assay standardisation as a means to generating assay data (not a strong driver per se). Focus on generating datasets from which to derive predictive algorithms. Shared data needs to be in a 'neutral' secure repository.
>Mid-term hopes: creation of large enough dataset (from which to derive predictive algorithms).
>Long-term hopes: implementation of database which can accrue over time.
>Risks (Challenges):
-Not willing to offer up details on in house assay expertise i.e. precise conditions for assays have taken time to develop.
Mitigation: need certainty that all companies in same position would share details.
-Projects never really die so regarding data from 'dead' projects this will be limited and would need verification before sharing.
Mitigation: due to large variability in sequences of Ab from screening, focus on those which are not close to molecules of interest yet still have associated developability data (MedI)
>Chris Lloyd (Deactivated): mostly agree with GSK view although identifying and sharing data from 'dead' projects not seen as so much of an issue (see Mitigation above).
>Bryan Jones (Deactivated): mostly agree with GSK and willingness to share some data on sequences.
>Francisco Fernandez (Deactivated): will share sequences if this is the agreed way forward.
'Basecase' scenario discussion
>Discussion suggests that there is lack of clarity and alignment regarding what the goals are, the scope and the approach to take.
>Questions and comments arising:
-Need to be clear on what we want to do with the data? - data comparison will be very hard if the dataset is heterogeneous i.e containing different Ab-types, so suggested focus is on IgG1's (needs to be made clear in scope).
-The above suggests the focus is on historical data, thus how are we planning to interpret the data given that the same assay will be carried out differently at different organisations? Otherwise the focus is on generation of new data based on assays put forward by organisations (perhaps in a standard format?) and then sharing this common 'pool' of data.
-What are we going to do that will differentiate us from the Jain et al (2017) paper, where they established 'ideal' assay ranges from a set of clinical Abs? In some ways the paper fell short of concluding which, if any, assays can be left out when running a standard campaign.
-Is there ways to identify new assays or combination of assay data to predict manufacturability of molecule? This implies collecting new data on the same set of molecules.
-How do we define value? is it in terms of how effectively we can pick out the molecules which will require more resources to develop or least resources to fix easily?
-If we are looking to establish a minimum set of 'value' assays then historical and new data are probably needed as we will need the entire range of data (early to endpoint) and not all molecules will have this.
-Will developing predictive algorithms based on the data be done as a group or individually? This will depend to some extent on how willing organisations are to share primary data.
>What we want to get out from establishing this database and would add value beyond the Jain et al (2017) paper: deliver added predictive power based on sharing existing data or generating new data to understand which are the minimum set of experiments needed to run a 'standard' campaign <still need to agree the boundaries> and understand the variability between experiments run a different organisations (to enable comparisons with internal data). Understanding what is driving undesired flags/properties from a structure/biophysics perspective.
>Next steps:
- State and agree our vision,
- Reach agreement on meaning of value and the value proposition,
- Agree our objectives,
- Discuss and agree the overall approach,
- Compile basecase scenario and present approach in an attractive way to internal and external stakeholders.