# Common Pitfalls in Replication ## Quick Answer - True replication means independent experimental units, not repeated measurements within one unit, and conflating the two invalidates statistical inference. - Design your experiment so each treatment is applied to multiple independent units, then record unit-level responses for analysis. - A key limitation is that technical replicates can improve precision but cannot substitute for biological replication when you aim to generalize. ## At a Glance | Concept | Common Mistake | Correct Approach | Consequence of Error | | --- | --- | --- | --- | | Replication | Increasing subsamples within one animal or culture | Use multiple independent animals or cultures per treatment | Inflated confidence and false significance | | Pseudoreplication | Treating repeated measures from one unit as independent | Analyze unit-level averages or use mixed models | Overstated degrees of freedom and invalid p-values | | Randomization | Assigning treatments without random allocation | Randomly assign treatments to experimental units | Confounded variables and biased estimates | | Controls | Omitting negative or positive controls | Include appropriate controls for each experiment | Inability to attribute effects to treatment | | Blinding | Assessing outcomes without masking treatment | Blind outcome assessment where feasible | Observer bias and inflated effect estimates | | Reporting | Omitting replication details in methods | Report unit definitions, sample sizes, and analysis methods | Irreproducible results and poor peer review | ## Understanding Replication in Experimental Design Replication is the independent application of a treatment to multiple experimental units. The experimental unit is the smallest entity to which a treatment is independently assigned. In life science research, this unit can be an animal, a plant, a cell culture flask, or a field plot. The key is that each unit receives the treatment independently and responds independently to it. A common error is to confuse technical replicates with biological replicates. Technical replicates are repeated measurements from the same experimental unit. Biological replicates are independent units that receive the same treatment. For example, measuring the same blood sample three times is technical replication, while measuring blood from three different animals is biological replication. Only biological replication supports inference about the population from which the units were drawn. The distinction matters because statistical tests assume independence among observations. When you treat technical replicates as independent, you artificially inflate the sample size and degrees of freedom. This leads to smaller p-values and narrower confidence intervals than the data actually support. The result is a false sense of certainty and conclusions that do not generalize. ## Core Principles of Experimental Design ### The Experimental Unit The experimental unit is the smallest entity that receives a treatment independently. In animal studies, the unit is usually the individual animal. In cell culture, the unit is the flask or well that receives a distinct treatment. In field trials, the unit is the plot. The choice of experimental unit determines how replication is counted. When you apply a treatment to a group of animals in one pen, the pen is the experimental unit, not the individual animals. If you then measure each animal in the pen, you have multiple measurements from one unit. These measurements are not independent because the animals share the same environment and treatment application. Analyzing them as independent observations is pseudoreplication. ### Independence and Randomization Randomization ensures that treatment assignment is not systematically biased by uncontrolled factors. Without randomization, differences between treatment groups may be due to preexisting differences instead of the treatment itself. Randomization also provides the basis for statistical tests that assume independent observations. Randomization should be applied at the level of the experimental unit. For example, if you have 20 animals and two treatments, you should randomly assign each animal to one of the two treatments. If you randomize cages instead of animals, the cage is the unit and the analysis must reflect that. ### Controls and Comparison A control group provides a baseline against which treatment effects are measured. The control should be identical to the treatment group except for the treatment itself. This allows you to attribute differences to the treatment instead of to other factors. Controls can be negative, meaning no treatment is applied, or positive, meaning a known treatment is applied for comparison. The choice of control depends on the research question. A negative control shows the baseline response, while a positive control shows the expected response to a known treatment. ## Common Failure Patterns ### Pseudoreplication Pseudoreplication occurs when the number of observations in the analysis exceeds the number of independent experimental units. This happens when multiple measurements are taken from the same unit and treated as independent. The result is an overstatement of the sample size and an understatement of the variability. A classic example is a study where one culture is treated and one culture is untreated, but multiple samples are taken from each culture. The analysis treats each sample as an independent observation, but the actual sample size is two, not the number of samples. This error leads to false confidence in the results. ### Confounding Confounding occurs when the effect of a treatment is mixed with the effect of another variable. This can happen when treatment groups differ in a factor other than the treatment. For example, if all treated animals are housed in one room and all controls in another, the room effect is confounded with the treatment effect. Randomization and blocking are used to reduce confounding. Blocking involves grouping experimental units into blocks that are similar, then randomly assigning treatments within each block. This controls for variation between blocks and allows for a more precise estimate of the treatment effect. ### Lack of Blinding Blinding is the practice of keeping the treatment assignment hidden from the researcher or the subject. This reduces bias in the assessment of outcomes. If the researcher knows which treatment a unit received, they may unconsciously influence the measurement or interpretation. Blinding is especially important when the outcome is subjective, such as a behavioral score or a histologic assessment. Blinding can be single, meaning the researcher is blinded, or double, meaning both the researcher and the subject are blinded. ## Practical Workflow for Designing a Replicated Experiment ### Step 1: Define the Experimental Unit Identify the smallest unit that receives the treatment independently. This is the unit of replication. For animal studies, it is the individual animal. For cell culture, it is the flask or well. For field studies, it is the plot. ### Step 2: Determine the Number of Replicates The number of replicates is the number of independent units per treatment. This number depends on the variability of the system and the size of the effect you want to detect. More replicates increase the power of the study to detect a true effect. ### Step 3: Randomize Treatment Assignment Randomly assign treatments to the experimental units. This can be done with a random number generator or a table of random numbers. The goal is to ensure that each unit has an equal chance of receiving each treatment. ### Step 4: Apply the Treatment Apply the treatment to each unit according to the assignment. Keep the treatment application consistent across all units. Record the date and time of application and any deviations from the protocol. ### Step 5: Measure the Outcome Measure the outcome for each unit. If multiple measurements are taken from the same unit, record them as repeated measures. The unit of analysis is the unit, not the individual measurement. ### Step 6: Analyze the Data Analyze the data at the level of the experimental unit. If you have multiple measurements per unit, you can average them or use a mixed model that accounts for the correlation. Do not treat repeated measurements as independent. ## Records and Measurements ### Documentation of Replication Record the number of experimental units per treatment and the number of measurements per unit. This information is essential for the analysis and for the interpretation of the results. It also allows other researchers to assess the validity of the conclusions. ### Data Recording Record the data in a structured format, such as a spreadsheet or a database. Include the treatment assignment, the unit identifier, and the outcome measurements. Keep the raw data and the analysis code for transparency. ### Quality Checks Check the data for errors, such as missing values, outliers, or inconsistent units. Verify that the treatment assignments match the protocol. Check that the measurements are within the expected range. ## Common Failure Patterns ### Failure to Randomize A common failure is to assign treatments in a systematic way, such as by order of arrival or by cage location. This can introduce bias if the systematic factor is related to the outcome. Randomization is the only way to avoid this bias. ### Failure to Blind Another failure is to assess the outcome without blinding. This can lead to bias in the measurement or interpretation. Blinding is especially important when the outcome is subjective or when the researcher has a strong expectation of the result. ### Failure to Report Replication A failure to report the number of replicates and the unit of analysis makes it impossible for the reader to assess the validity of the results. Reporting guidelines, such as those from the EQUATOR Network, provide a framework for transparent reporting of experimental design. ## Options and Tradeoffs ### Increasing Replicates Increasing the number of replicates increases the power of the study to detect a true effect. However, it also increases the cost and the time required. The tradeoff is between the precision of the estimate and the resources available. ### Using Mixed Models Mixed models allow for the analysis of repeated measurements from the same unit. They account for the correlation between measurements and provide valid inference. The tradeoff is the complexity of the analysis and the need for specialized software. ### Pooling Samples Pooling samples from multiple units can reduce the cost of measurement. However, pooling loses information about the variability between units. The tradeoff is between the cost of the measurement and the precision of the estimate. ## Observations and Measurements ### Recording the Unit of Analysis The unit of analysis should be recorded in the methods section of any report. This includes the number of units per treatment and the number of measurements per unit. This information is essential for the reader to evaluate the validity of the results. ### Recording the Randomization Scheme The randomization scheme should be recorded, including the method used to generate the random assignment. This allows the reader to verify that the assignment was truly random. ### Recording the Blinding Procedure The blinding procedure should be recorded, including who was blinded and how the blinding was maintained. This is important for the reader to assess the risk of bias. ## Quality and Welfare Controls ### Animal Welfare In animal studies, the number of replicates should be the minimum necessary to achieve the scientific objective. This is a requirement of the 3Rs principle, which stands for Replacement, Reduction, and Refinement. The number of animals should be justified in the protocol. ### Data Quality The quality of the data depends on the accuracy of the measurements and the consistency of the treatment application. Standard operating procedures should be followed to ensure consistency. The data should be checked for errors and outliers. ### Biosafety The treatment application should follow biosafety guidelines to protect the researcher and the environment. This includes the use of personal protective equipment and the proper disposal of waste. ## Limitations and Interpretation ### Limits of Replication Replication does not guarantee that the results are true. It only provides an estimate of the variability of the effect. The results are still subject to bias and confounding. ### Limits of Statistical Tests Statistical tests are based on assumptions about the data, such as normality and independence. If these assumptions are violated, the results of the test may be invalid. The data should be checked for these assumptions before the test is applied. ### Limits of Generalization The results of a study apply to the population from which the sample was drawn. If the sample is not representative of the population, the results may not generalize. The limitations of the study should be stated in the report. ## Safety and Regulatory Context ### Institutional Oversight Institutional review boards and animal care committees oversee the conduct of research. They review the protocol to ensure that the number of replicates is justified and that the welfare of the animals is protected. ### Data Management The data management and sharing policy of the National Institutes of Health requires that data be shared in a timely manner. The data should be stored in a repository and made available to other researchers. ### Publication Ethics The Committee on Publication Ethics provides guidance on the reporting of research. This includes the requirement to report the number of replicates and the unit of analysis. The guidance also covers authorship, peer review, and data integrity. ## Professional Escalation Criteria ### When to Consult a Statistician A statistician should be consulted when the design is complex or when the analysis is not straightforward. This includes the use of mixed models, the analysis of repeated measures, or the design of a study with multiple factors. ### When to Consult an Ethics Committee An ethics committee should be consulted when the study involves animals or human subjects. The committee reviews the protocol to ensure that the welfare of the subjects is protected and that the number of replicates is justified. ### When to Consult a Biosafety Officer A biosafety officer should be consulted when the study involves hazardous materials or organisms. The officer reviews the protocol to ensure that the safety of the researcher and the environment is protected. ## A Decision Framework for Matching Replication to Your Research Question The most persistent replication errors do not come from a lack of awareness that independent units matter. They come from a mismatch between the research question and the level at which replication is executed. A researcher who asks whether a drug changes gene expression in a tumor might measure three samples from one tumor and call that replication. Another researcher who asks whether a housing system affects weight gain might house ten animals per pen, treat the pen as the unit, and then wonder why the analysis has so little power. Both are making the same error in different directions: the level of replication does not match the level at which the question is asked. This section provides a practical decision framework that forces you to state your research question in terms of the population you want to make claims about, then work backward to identify the experimental unit, the number of independent units needed, and the analysis that matches that structure. The framework is designed to be used before data collection begins, and it produces a written record that can be included in a protocol, a grant application, or a methods section. ### Step 1: State the Target Population in Plain Language The first step is to write down the population to which you intend to generalize. This is not the same as the sample you will measure. It is the larger set of units that your results should apply to. For example, if you are studying a feed additive, the target population might be "commercial broiler flocks raised under standard industry conditions." If you are studying a cell line, the target population might be "all cultures of this cell line under the same passage and growth conditions." If you are studying a surgical technique, the target population might be "patients with this condition who would receive this surgery." Write this statement in one sentence. If you cannot write it, you cannot determine the correct level of replication. The target population statement forces you to be explicit about the scope of your inference. It also prevents the common error of designing a study that is only valid for the specific animals or cultures in your laboratory. ### Step 2: Identify the Experimental Unit The experimental unit is the smallest entity that receives the treatment independently and that you would expect to respond independently. This is the entity that you will replicate. The definition is not always obvious, and it depends on how the treatment is applied. Ask this question: what is the smallest entity to which I can assign a treatment and observe a response that is not shared with another entity? If you apply a treatment to an individual animal, the animal is the unit. If you apply a treatment to a pen of animals, the pen is the unit. If you apply a treatment to a flask of cells, the flask is the unit. If you apply a treatment to a well in a plate, the well is the unit. The most common error is to choose a unit that is smaller than the level at which the treatment is applied. For example, if you treat a single culture flask and then split the culture into three wells for measurement, the flask is the unit, not the wells. The three wells are technical replicates. They tell you about the precision of your measurement, not about the variability between independent cultures. The second most common error is to choose a unit that is larger than the level at which the treatment is applied. For example, if you treat individual animals but house them in groups, the individual animal is the unit, but the housing group is a source of shared variation. This is a blocking factor, not the experimental unit. The analysis must account for the shared housing. ### Step 3: Count the Number of Independent Replicates Once the experimental unit is identified, count the number of units that receive each treatment. This is the number of replicates. The number of measurements you take from each unit does not change this count. A useful test is to ask: If I remove one unit from the treatment group, does the treatment still have any other units that received it independently? If the answer is no, you have one replicate. If the answer is yes, you have that many replicates. For example, if you have three culture flasks that all received the same treatment, you have three replicates. If you have one flask and you measure it three times, you have one replicate. If you have one pen of ten animals, you have one replicate. If you have three pens of ten animals, you have three replicates. The number of replicates determines the precision of your estimate and the power of your statistical test. More replicates give you a more precise estimate of the treatment effect and a better chance of detecting a true effect. Fewer replicates give you a less precise estimate and a higher chance of missing a true effect. ### Step 4: Match the Analysis to the Unit The analysis must treat the experimental unit as the independent observation. If you have one measurement per unit, the analysis is straightforward. If you have multiple measurements per unit, you have two options. The first option is to average the measurements within each unit and analyze the unit-level averages. This is simple and valid. The downside is that you lose information about the variability within units, but this variability is not what you are trying to estimate. The second option is to use a mixed model that includes the unit as a random effect. This allows you to use all the measurements while accounting for the correlation between measurements from the same unit. This is more complex and requires specialized software, but it can be more efficient when the number of measurements per unit is large. The choice between these two options depends on the research question and the data structure. If the within-unit variability is small, the unit-level average is often sufficient. If the within-unit variability is large, a mixed model may be more appropriate. A statistician can help you decide. ### Step 5: Write the Replication Statement The final step is to write a replication statement that describes the design in a way that another researcher can understand. This statement should include the target population, the experimental unit, the number of replicates per treatment, and the analysis method. A template for the statement is: "The target population is [population]. The experimental unit is [unit]. Each treatment was applied to [number] independent units. The outcome was measured [number] times per unit. The data were analyzed at the level of the [unit] using [analysis method]." This statement should be written before data collection begins. It serves as a check on the design and as a record for the methods section. It also helps you identify problems before you spend time and resources on an experiment that cannot answer the question. ## A Record System for Replication Decisions The decision framework is only useful if it is recorded and used consistently. A simple record system can prevent the most common replication errors. The system has three parts: a design log, a unit map, and an analysis note. ### Design Log The design log is a dated entry that records the research question, the target population, the experimental unit, the number of replicates, and the randomization scheme. It is written before data collection and updated if the design changes. The design log is the first place to look when a problem arises in the analysis. The design log should include the following fields: - Date of design - Research question - Target population - Experimental unit - Number of replicates per treatment - Randomization method - Blinding method - Any deviations from the plan ### Unit Map The unit map is a diagram or table that shows the relationship between the experimental units and the measurements. It is a visual representation of the design. The unit map should show each unit, the treatment it received, and the measurements taken from it. The unit map is useful for identifying pseudoreplication. If the map shows multiple measurements from the same unit, you can see that the unit is the analysis level. If the map shows one measurement per unit, the analysis is straightforward. ### Analysis Note The analysis note records the statistical method used and the level at which the analysis was performed. It is written after the data are collected and before the analysis is run. The analysis note should state whether the unit-level averages were used or whether a mixed model was used. It should also state the software and the version. The analysis note is useful for the methods section of a paper. It provides a record of the analysis that can be reported transparently. It also helps you avoid the error of analyzing the data at the wrong level. ## Troubleshooting a Design That Is Already in Progress The decision framework is best used before data collection, but it can also be used to troubleshoot a design that is already in progress. If you have already collected data and you are not sure whether the replication is correct, you can use the framework to diagnose the problem. ### Diagnosis 1: The Unit Is Too Small If you treated one culture and measured it multiple times, the unit is the culture, not the measurement. The analysis must use the culture as the unit. If you have only one culture, you have one replicate. You cannot make a statistical claim about the effect of the treatment because you have no estimate of the variability between cultures. The solution is to collect more independent units. If that is not possible, you must limit your conclusions to the specific culture you studied. You cannot generalize to other cultures. ### Diagnosis 2: The Unit Is Too Large If you treated a group of animals in one pen and measured each animal, the unit is the pen, not the animal. The analysis must use the pen as the unit. If you have one pen, you have one replicate. The measurements from the animals in the pen are not independent. The solution is to use the pen-level average as the outcome. If you have only one pen, you cannot make a statistical claim about the treatment. ### Diagnosis 3: The Unit Is Confounded If the treatment is applied to one group of animals in one room and the control is applied to another group in another room, the room is confounded with the treatment. You cannot separate the effect of the treatment from the effect of the room. The solution is to repeat the experiment in multiple rooms or to use a randomized block design. If the data are already collected, you can only report the limitation. ## Common Failure Patterns in the Decision Framework The decision framework is simple, but it is easy to make mistakes. The following are the most common failure patterns. ### Failure to State the Target Population The most common failure is to skip the first step and go directly to the design. Without a target population statement, the researcher does not know what the replication is for. The result is a design that is either too small or too large for the question. ### Failure to Identify the Unit Correctly The second most common failure is to identify the unit incorrectly. This happens when the treatment is applied at one level and the measurement is taken at another level. The researcher must be explicit about the level at which the treatment is applied. ### Failure to Count Replicates Correctly The third most common failure is to count the number of replicates incorrectly. This happens when the researcher counts the number of measurements instead of the number of units. The number of replicates is the number of independent units, not the number of measurements. ### Failure to Match the Analysis to the Unit The fourth most common failure is to analyze the data at the wrong level. This happens when the researcher treats the measurements as independent when they are not. The analysis must use the unit as the observation. ## Practical Implementation Steps The decision framework is most useful when it is used as a checklist before data collection. The following steps are a practical way to implement the framework. ### Step 1: Write the Target Population Statement Write the target population statement in one sentence. This is the population you want to make conclusions about. If you cannot write this statement, you cannot design the experiment. ### Step 2: Identify the Experimental Unit Identify the smallest entity that receives the treatment independently. Write this in the design log. ### Step 3: Count the Replicates Count the number of independent units per treatment. Write this in the design log. ### Step 4: Plan the Analysis Plan the analysis at the level of the unit. Decide whether you will use unit-level averages or a mixed model. Write this in the analysis note. ### Step 5: Review the Design Review the design with a colleague or a statistician. Ask them to identify the unit and the number of replicates. If they identify a different unit than you did, the design is not clear. ## When to Escalate to a Statistician The decision framework is designed to be used by researchers, but there are situations where a statistician should be consulted. These situations include: - The design has multiple factors or multiple levels of nesting. - The outcome is measured at multiple time points. - The data are not normally distributed. - The design is unbalanced, meaning the number of replicates is not equal across treatments. - The researcher is not sure about the unit or the analysis. A statistician can help you identify the unit, determine the number of replicates, and choose the analysis. The statistician can also help you write the analysis note and the methods section. ## The Role of Reporting Guidelines The decision framework is a design tool, but it is also a reporting tool. The EQUATOR Network provides reporting guidelines that help researchers report the design and the analysis of their study. These guidelines include the ARRIVE guidelines for animal research and the CONSORT guidelines for clinical trials. The guidelines require the researcher to report the unit of analysis, the number of replicates, and the randomization scheme. Using the decision framework before data collection makes it easier to report the design in the methods section. The design log and the analysis note provide the information that the reporting guidelines require. The framework also helps you avoid the common failure of omitting replication details from the methods section. ## Limitations of the Decision Framework The decision framework is a tool, not a guarantee. It does not solve all design problems. The framework does not tell you how many replicates you need. That depends on the variability of the system and the size of the effect you want to detect. The framework does not tell you how to handle missing data or outliers. These are analysis problems that require statistical judgment. The framework also does not address the ethical and welfare considerations of the design. The number of replicates should be the minimum necessary to achieve the scientific objective. This is a requirement of the 3Rs principle, which stands for Replacement, Reduction, and Refinement. The framework helps you avoid the error of using too few replicates, but it does not help you avoid the error of using too many. ## Professional Escalation Criteria The decision framework is a self-audit tool, but it is not a substitute for professional judgment. You should escalate to a statistician when the design is complex or when the analysis is not straightforward. You should escalate to an ethics committee when the study involves animals or human subjects. You should escalate to a biosafety officer when the study involves hazardous materials. The decision framework is a practical tool that can be used by any researcher. It is not a replacement for statistical expertise, but it is a way to avoid the most common replication errors. The framework is a record system that can be used to document the design and the analysis. It is a troubleshooting method that can be used to diagnose a design that is already in progress. It is a comparison tool that can be used to evaluate the design of a study that is being reviewed. ## Frequently Asked Questions ### What is the difference between a technical replicate and a biological replicate? A technical replicate is a repeated measurement from the same experimental unit. A biological replicate is a measurement from a different experimental unit. Technical replicates measure the precision of the measurement, while biological replicates measure the variability of the effect. ### How many replicates do I need for my experiment? The number of replicates depends on the variability of the effect and the size of the effect you want to detect. A statistician can help you determine the number of replicates needed to achieve the desired power. ### What is pseudoreplication and why is it a problem? Pseudoreplication is the analysis of data as if the observations are independent when they are not. This happens when multiple measurements are taken from the same unit and treated as independent. It leads to an overstatement of the sample size and an inflated confidence in the results. ### How do I randomize the treatment assignment? Randomization can be done with a random number generator or a table of random numbers. The goal is to ensure that each unit has an equal chance of receiving each treatment. The randomization scheme should be recorded. ### What is the role of a control group? A control group provides a baseline for the comparison of the treatment effect. The control group is identical to the treatment group except for the treatment itself. This allows the researcher to attribute differences to the treatment. ### How do I report the replication in my paper? The methods section should report the number of experimental units per treatment and the number of measurements per unit. The unit of analysis should be stated. The randomization scheme and the blinding procedure should also be reported. ### What is the role of the EQUATOR Network? The EQUATOR Network provides reporting guidelines for the transparent reporting of research. These guidelines help researchers report the design and the analysis of their study in a way that is clear and complete. ### What is the NIH Data Management and Sharing Policy? The NIH Data Management and Sharing Policy requires that data be shared in a manner that is consistent with the principles of transparency and reproducibility. The policy requires a data management plan and the sharing of data in a repository. ## Using the Evidence | Source | Best use in this topic | Important limitation | |---|---|---| | [Research Methods Resources](https://www.ncbi.nlm.nih.gov/books) | official guidance | Check the linked page for current local requirements | | [EQUATOR Network](https://www.equator-network.org/) | official guidance | Check the linked page for current local requirements | | [Core Practices](https://publicationethics.org/core-practices) | official guidance | Check the linked page for current local requirements | ## Related Diagnostic Guides - [Multiplexing in Sequencing: How to Pool Samples Efficiently](/knowledge/diagnostics/molecular/multiplexing-in-sequencing-how-to-pool-samples-efficiently) - [Animal Biology](/blog/guides/animal-biology) - [Animal Research](/blog/guides/animal-research) ## References and Further Reading - [Research Methods Resources](https://www.ncbi.nlm.nih.gov/books). National Library of Medicine. - [EQUATOR Network](https://www.equator-network.org/). EQUATOR Network. - [Core Practices](https://publicationethics.org/core-practices). Committee on Publication Ethics. - [NIH Grants and Funding](https://grants.nih.gov/). National Institutes of Health. - [ORCID for Researchers](https://info.orcid.org/researchers). ORCID. - [Data Management and Sharing Policy](https://sharing.nih.gov/data-management-and-sharing-policy). National Institutes of Health. - [Laboratory Biosafety Manual](https://www.who.int/publications/i/item/9789240011311). World Health Organization. - [NCBI Literature Resources](https://www.ncbi.nlm.nih.gov/). National Center for Biotechnology Information. - [Self-Replicating RNA.](https://pubmed.ncbi.nlm.nih.gov/27987141). Methods in molecular biology (Clifton, N.J.), 2017. - [Design and Implementation of Observational Studies Emulating a Target Trial.](https://pubmed.ncbi.nlm.nih.gov/41712213). JAMA network open, 2026. - [25th Annual Computational Neuroscience Meeting: CNS-2016.](https://pubmed.ncbi.nlm.nih.gov/27534393). BMC neuroscience, 2016. > This article is educational and does not replace validated laboratory procedures, institutional biosafety review, manufacturer instructions, or professional interpretation.