Most companies use established platform processes and perform minimal optimization before IND.

The focus is on generating sufficient material for early clinical studies.

Major optimization is typically performed later, after Phase I, once dose-ranging and mechanism-of-action data are available

The  core requirements remain the same: quality, regulatory compliance, and robustness. However, experience with specific molecule formats becomes much more important. Developers  should assess a provider's track record with: 

• Bispecific antibodies
• Multispecifics
• Fc-fusions
• Other complex biologics

AI and machine learning are used for early clone selection. Predictive modelling helps identify promising clones before extensive screening.

Data analytics and model-based approaches support later-stage clone evaluation and process development.

AI-based modelling can predict perfusion performance using fed-batch data.

Modelling can support process optimization by assessing:

• Feeding strategies
• Temperature shifts
• Process parameters

The goal is a smoother, more efficient transition to intensified or perfusion processes.

Three primary criteria:

• Titer/Productivity
• Product Quality and Critical Quality Attributes (CQAs)
• Clone Stability

Additional considerations:

Biosimilars require close matching to the originator's quality profile.

Novel biologics offer more flexibility but still require strong quality and efficacy data.

Increasing analytical data availability can make final clone selection more complex.

For complex molecules such as bispecifics, correct chain assembly and functionality become critical selection criteria.