...
- Laura hook (Deactivated)
- Bryan Jones (Deactivated)
- Chris Lloyd (Deactivated)
- Carmen Nitsche (Deactivated)
- Kieran Todd (Deactivated)
Discussion items
| Time | Item | Who | Output |
|---|---|---|---|
| 17:00 - 18:00 (60 mins) | >Next steps discussion (focus will be on 3 & 4).
>Upcoming discussions with Pfizer and Lonza | All | >Clarify, agree and capture on wiki points 1-5. >Team input on proposed approach. |
| If time permits | >EMBL-EBI presentation | >Team input on proposed presentation. |
Minutes
Next steps discussion
>Vision, Value proposition and Objectives agreed and captured in wiki during meeting.
>Problem statements are a reflection of the issues faced by industry partners and as written these suggest that there are three main needs to achieve the value proposition; (1) more biophysical and more endpoint data, (2) ability to predict endpoint data from biophysical data (3) a standard way to measure data to enable comparisions between different partners.
>Key point re. (3), if we are confident that we can compare between different assays and institutions, we can do without standardisation of assays. This may hold in some cases but not others where even a change in buffer conditions may lead to the same Ab flagging as having a risk. Intermediate goal is to highlight which assays are most reproducible and robust (possess least variability) as these are most likely to allow early comparison of data across industry partners. This approach is a step beyond what was done in the Jain et al (2017) paper since all the data generation is done in one place.
>Re. (1) and (2), biggest need is in endpoint data sharing and ability to compare may not hold for endpoint data as for early biophysical data. One good example of this is for viscosity, where dependency on the formulation is critical and where considerable differences would preclude direct comparisons (one way round may be to use R/A/G buckets rather than direct comparisons of the numbers). Need endpoint for both well behaved and poorly behaved molecules. Question: is there enough endpoint data available for poor molecules? Some data will have to be generated but combining internal data with already published data (including from Jain et al 2017) should be enough to get us started.
>The proposed data sharing is an opportunity to better define structural descriptors. Chris Lloyd (Deactivated) and Bryan Jones (Deactivated): there is internal evidence that SAP/SCM holds for ranking Abs of the same lineage but the correlation does not hold across other antibodies.