" It is not the strongest of the species that survives, not the most intelligent, but the one most responsive to change."
Charles Darwin

Optimizing High Throughput Life Science Research


Competitive pressures have forced both the not-for-profit and the for-profit life science research communities to embrace high throughput research. While high throughput processes have dramatically increased the pace of life science research, they have also dramatically increased the complexity of research processes. As these processes have grown more complex, they have also grown more sensitive to even small process changes, making process optimization and process yield management essential to remain competitive.

BergenShaw International’s
Focus high throughput research process optimization software enables gene sequencing, gene expression, genotyping, proteomics, biomolecular screening, and other high throughput research laboratories to optimize their processes, increase their yield, improve their throughput, and lower their overall operating cost. Focus analyzes the performance of all individual factors (libraries, machines, personnel, reagents, etc.) and factor sets (Library=Human7 and Machine=Thermocycler41 and Operator=JTS) associated with the process on an hourly or daily basis and pinpoints the factors and factor sets associated with all yield losses and gains.  Once Focus is configured, it runs automatically and notifies appropriate personnel of performance anomalies based on user specified criteria.  Focus enables high throughput research laboratories to not only survive but to thrive by accelerating their response to change.

 

Yield – A simple concept with a complex reality

Research process yield is a simple concept.  The diagram below represents the process flow through a typical high throughput gene sequencing laboratory.  Samples are transformed through ten process steps into DNA sequence chromatograms.  Each chromatogram is then processed by a base calling program such as Phred and assigned a quality value.  Sequencing attempts that meet an established quality threshold are a success.  Those that do not are a failure.  The yield for this example process is simply the ratio of the number of successful sequencing attempts to the total number of sequencing attempts for a given period.

Although yield is simple in concept, it has a complex reality.   The diagram below provides an example of the complexity inherent in any high through research process.  Samples are likely to come from many different sources.  Process step throughput rate incompatibilities require that each step be performed by a number of different work cells.  All of the work cells in any given process step may not use the same make and model of equipment or batch or manufacturer of reagents.  Add to this, shift-to-shift and operator-to-operator differences and the many other factors that affect overall process performance and the simple concept of yield displays a very complex reality.

 

Eliminating Yield Loss

Overall yield loss is made up of a number of specific individual yield losses, each with a unique cause.  The cause of simple, single factor induced losses, such as when the temperature of a single machine is outside of its specified operating range, can easily be identified by SPC (Statistical Process Control).  The comparative yield performance of members of a single resource type, such as the electrophoresis instruments in a gene sequencing process, can be determined by POV (Partition Of Variance) or other ANOVA (Analysis Of Variance) techniques.  However, these techniques are of little use in the extremely difficult task of identifying the specific factors involved in a multi-factor-induced loss where two or more process factors interact or “conspire” to cause a specific yield loss.  For example, a temperature on a thermocycler at one process step interacts with a reagent used in the next process step, and an electrical field produced by an instrument several steps later in the process to produce an unexpected yield change.  In complex high throughput research processes, the interactions of even “within specified operating range” process factors are far too complicated to predict the “emergence” of such yield changes.

In the past, process managers have relied on DOE (Design Of Experiments) to identify the factor set associated with a multi-factor induced yield loss.  However, using DOE requires that an expert in the process under study hypothesize theories of cause and effect for each yield loss, then design, precisely execute, and rigorously analyze the results of experimental process scenarios to test the accuracy of each hypothesis.  DOE often requires a number of experimental cycles before the process factors associated with a specific yield loss can be identified.  Due to the time and cost associated with the experimental process, DOE is usually reserved for only the largest individual yield losses.  Smaller individual yield losses, aggregated together, often represent a significant portion of overall yield loss, but are usually not viewed as practical DOE candidates.

DOE is hypothesis driven.  Focus, on the other hand, is discovery driven.  Focus does not require an hypothesis.  Focus treats each sample run through a research process as an experiment and uses proprietary algorithms to test all possible hypotheses by those experiments.



Focus – High Throughput Research Process Optimization


BergenShaw International’s Focus high throughput research process optimization software rapidly identifies the specific process factors associated with every individual cause of yield loss, reducing from days to minutes the time required to identify the factors associated with any specific yield loss.  Focus also identifies the factors associated with every yield gain enabling unexpected process improvements to be identified and capitalized on. 

Returning to the generic high throughput research process example used above, a sample may take the path through that process shown in blue in the diagram below. At each process step a sample tracking or laboratory information management system would collect data about the factors associated with that step at the time the sample was being processed. These factors could include instrument manufacturer, instrument model, instrument ID, sample hold time from the previous process step, date, time, shift, operator(s), reagent manufacturer(s), reagent lot numbers, protocols, pressures, temperatures, flow rates, voltages, currents, hours since the equipment was last serviced, number of samples processed since the equipment was last serviced, technician(s) last servicing the equipment, and any other relevant factors.


Focus compares the performance with every process factor, individually and in combination with every other process factor associated with every process step providing a comparison between the performance of individual factors and groups of factors or factor sets in the process overall.  In addition, Focus calculates the impact on the number of samples successfully processed based on the performance of each factor set.  The data contained on the first line of the segment of a Focus Results Table shown below indicates that for the 13,248 samples that where processed through (associated with) both Amplifier Cycler 1229 and Cycle Sequence Cycler 192 the Yield loss was 2.92 times the target yield loss which resulted in the loss of an additional 1275 samples.  The number in parentheses around each of the individual factors in a multi-factor Factor Set indicates the Relative Yield Loss for that factor alone.  As is evident from the data on this line, the interaction between two poorly performing factors with a relative yield loss of 1.75 (75% worse than the target yield loss) and 1.53 (53% worse than the target yield loss) respectively, created a much worse performing Factor Set with a Relative Yield Loss of 2.92 (192% worse than the target yield loss).  However, even though the first line’s Factor Set was the worst performing of all of those show, each of the other three Factor Sets had a greater impact on the number of samples lost do to their association with a greater number of samples overall.  This is because Impact is based on Relative Yield Loss AND Units In.

The segment of the Focus Results Table shown below is from the same analysis as the one above and shows some of the best performing Factor Sets analyzed.  While most Aliquot Primer Lots had a Relative Yield Loss of between .90 and 1.10 in this analysis, Aliquot Primer Lot performed significantly better with a Relative Yield Loss of .70, 30% better than the target yield loss.  Such unexpected yield gain may warrant further investigation as they may represent process enhancement opportunities.

 

 

To facilitate further study, Focus can produce trend charts, loss category charts, histograms, summary charts, and source data statistics reports based on any individual factor set or group of factor sets.

Since most high throughput life science research is done using 96, 384, or 1536 well microtiter trays or some other multi-unit carrier to facilitate automation, Focus can analyze both tray/plate based data and well/sample based data.  When analyzing well/sample data, Focus can produce well charts graphically illustrating the performance of each individual well location.  Well charts allow for the rapid identification of well related problems with automation systems and plate region (edge, corner, center, etc.) problems. 

Focus is a Microsoft Windows client server application. Well/sample or tray/plate process history records are transferred hourly or daily from the sample tracking system or laboratory information system server to the Focus server. The Focus server automatically runs user defined analysis, charts analysis results, and notifies the specified users by e-mail and/or pager when specific user defined conditions are present in the analysis results.

Focus client software is used to define the analysis, charting, and notification parameters to be run automatically and to perform ad hoc analysis and charting.

Summary

Much of life science research today is based on high throughput research processes.  The key to successful high throughput research is to optimize research process throughput.  BergenShaw International’s Focus software product is the world’s most advanced high throughput research process optimization tool.  Focus enables high throughput laboratories to optimize their research processes, thus increasing their yield, improving their throughput, and lowering their overall operating cost. Once configured, Focus automatically enables your laboratory to be the one most responsive to change.