How to achieve high throughput in biopharma analytical workflows

High throughput is often treated as a hardware question.

• How many samples can the system run per hour?
• How many plates, vials or positions can it handle unattended?
• How long is the measurement cycle?

All fair questions, and all about the hardware. But none of these questions asks how fast a lab can actually use the results for decision making.

What is “end-to-end workflow throughput” in a biopharma analytical process?

The end-to-end workflow throughput is the number of end-point results analytical teams access in a unit of time.

Here, we’re talking about end-point results that can be used for decision making, not just data that still needs to be processed or organised.

In fact, we’re not talking about the throughput of an instrument, but we’re rather considering the whole workflow: sample analytics + data analytics.

Here’s how they’re different:

• Instrument throughput: number of samples processed per hour (or day)
• End-to-end workflow throughput: number of end-point results accessed per hour (or day)

This concept is a more accurate reflection of ‘high throughput’ as it tells you how fast analytical teams receive results that they can use for decision making.

As an example; if you analyse 384 samples per hour, but then require additional hours of time to interpret or organise that data or troubleshoot outliers, the end-to-end workflow throughput may not be so favorable.

Especially in busy labs, the gap between instrument throughput and workflow throughput can be significant.

The more samples a lab runs, the more data, metadata, exceptions and review tasks it creates. If those outputs are not structured, traceable and easy to interpret, higher instrument throughput can move the operational bottleneck from the instrument to the analyst, reviewer or data scientist. The lab gets more data, but not necessarily more usable results they can base their decisions on rapidly.

“We have been innovating analytical workflows in biopharma for years. The biggest misconception we see is the perceived high throughput which reflects the instrument throughput only and does not consider the end-to-end perspective: from sample to usable data.”

Dominik Sievert, CEO of CLADE

What is first-pass data usability for biopharma analytics

First-pass data usability is a metric that says whether a result can be reviewed, trusted and used directly, at a first-pass (without the need for manual data evaluation or organisation).

This may mean that:

• It has the correct sample identity.
• It carries the method or model version.
• It is linked to the right batch, formulation, process step or study.
• It includes the right units, flags and acceptance context.
• It is stored where the next person or system can use it.

First-pass data usability is a useful metric to consider when evaluating end-to-end throughput of an analytical workflow.

End-to-end throughput = samples measured x first-pass data usability

A workflow that runs 200 samples per day but requires extensive manual data cleanup may be less efficient than a workflow that runs fewer samples but produces decision-ready data.

This is especially true in biopharma, where analytical results feed formulation decisions, process development models, stability assessments, release testing packages and regulatory documentation.

An example of an end-to-end high throughput analytical workflow

In order to achieve end-to-end high throughput, we require an integrated workflow bringing together sample analytics and data analytics.

The bioanalytics solutions from CLADE integrate:

• Hardware – MIRA Analyzer performs automated measurements of samples
• Software – the data analytics tool Sphere and proprietary algorithms take care of data processing

As the software ecosystem includes a comprehensive infrastructure driven by data science, an analyst can get end-point results in 4 minutes. Here, we’re not talking about data that needs to be cleaned or processed, but rather end point results that inform decision making.

Learn more about this setup: