Best practices for using biological data in psychological research☆ Authors: Sage E. Hawn, Sinead M. Sinnott, Alicia K. Smith, Kyle Bourassa, Leslie Brick, Sian Hemmings, Soraya Seedat, Arash Javanbakht, Lauren Smith, Antonia V. Seligowski, Nathan A. Kimbrel, Divya Mehta, Hossein Abbasi, Erika Wolf Published: 2026 ABSTRACT Advances in statistical techniques for analyzing biological data have driven the need for larger sample sizes, which are ultimately made possible through team science efforts. Findings from these efforts have led to a significant shift in approaches to biological data collection over the past decade, from traditional approaches relying on single measures or assays collected by individual investigators in small lab-based studies to high-dimensional methodologies embedded within large cohorts or multi-site studies. These recent large scale consortium efforts necessitate careful consideration across the full data pipeline, from pre-data generation and sample collection to processing and statistical analysis, while addressing challenges such as cost, standardization of laboratory procedures, and harmonization across sites. Selecting an appropriate methodological approach to measure a biomarker involves complex decisions and tradeoffs to determine the most suitable biological modality (e.g., blood vs. saliva) and assay type (e.g., plasma vs. serum) for the research question and target population. These decisions must account for contextual factors such as cost, study population characteristics, storage space, and analytic capabilities, and carry important implications for data quality, feasibility, and interpretability. Standardization of data collection, storage, quality control, and analytical practices (e.g., accounting for technical confounders), is critical, especially when analyzing biological data across multiple cohorts and time points. This review synthesizes recommended best practices for standardizing collection of peripheral biomarkers (blood, saliva, and urine) in psychiatric research, which we hope will enhance data quality, facilitate cross-study comparisons, and advance individual and team biomarker research now and in the future. Introduction Psychiatric research is increasingly embracing the integration of biological data to advance understanding of mental health. Biobehavioral data can be used to study biological risk for psychopathology (e.g., genomic variation), as well as its correlates and consequences (e.g., immune dysregulation). Biomarkers are objectively measured indicators of biological processes that provide information about normal biological functions, disease-related changes, or responses to exposures or therapeutic interventions [1]. They can reflect a wide range of biological activity, including molecular, neurobiological, cellular, physiological, or systemic changes. Accordingly, biomarkers have the potential to deepen our understanding of the complex interplay between psychological and physical disease states, can be used to identify risk for a range of negative outcomes, and are crucial to the development and tailoring of psychotropic medications. Biomarkers also have the potential to serve as useful diagnostic tools, offering an objective solution for mitigating the ambiguity that plagues psychiatric nosology. The integration of biomarkers into psychiatric research reflects a growing recognition of how mental and physical health operate in concert to shape overall disease risk. However, methodological inconsistencies across studies pose significant challenges, limiting the comparability and reproducibility of findings and creating barriers for collaborative team science initiatives. We aim to provide a brief contextual overview of the evolution of the use of biomarkers in psychological research and identify best practices for standardizing the collection and analysis of biomarkers derived from peripheral biofluids (i.e., blood, saliva, and urine) for collaborative research efforts. We limit our review to minimally invasive biomarkers that can be assessed in living humans and that are applicable to a broad range of disciplines. As such, this review will not include recommendations for collecting and analyzing imaging biomarkers, nor biomarkers derived from postmortem tissue or stem cells, as these approaches often require different methodological, ethical, and scientific considerations, and may be less applicable to broader audiences and large population analyses. Historical overview of biomarkers in psychological research Over the past several decades, biobehavioral research has undergone a significant shift—from traditional approaches relying on single measures or assays collected by individual investigators in small lab-based studies, to high-dimensional methodologies that may be embedded within larger cohort studies. Until the 1990s, most research relied on correlational or observational data to explore associations between biology and psychopathology. In 1991, Cohen and colleagues published the first experimental study in humans demonstrating a dose dependent relationship between psychological stress and incident development of infection from common viruses [2]. This and similar work spurred a body of research focused on biological mechanisms underlying the connection between psychological stress and physical health and catapulted the use of biomarkers in psychiatric research. The burgeoning integration of biomarkers into the psychological sciences in the 1990s coincided with the establishment of large cohort panel studies, such as the Health and Retirement Study (HRS) in 1992 [3] and the Midlife in the United States (MIDUS) Study in 1994 [4]. Cohort studies such as these sought to provide large, representative samples that could support research across a wide range of scientific disciplines. In the early 2000’s, these studies began to expand beyond survey panels to include biomarkers. As biomarkers became integrated into large panel studies, it became feasible to explore longitudinal associations between psychiatric traits and health-related biomarkers among large, well-powered samples, something that had previously been practically impossible for any individual investigator. Such work has illuminated the dynamic associations between physiological states and mood and anxiety disorders. To demonstrate, data from the English Longitudinal Study of Ageing showed that depressive symptoms were significantly associated with cortisol and inflammation, measured via C-reactive protein levels, across a 14-year study period [5]. Another study presented data from two large, longitudinal cohorts demonstrating that higher levels of systematic inflammation, measured by interleukin−6 (IL-6), were associated with worse depression symptom trajectories across the lifespan [6]. Taken together, these studies underscore the utility of longitudinal biomarker research by suggesting that peripheral biomarkers may reflect both antecedents and consequences of mood disorders. Although an exciting development, large cohort studies have historically included biomarkers that can be collected and analyzed in many individuals at a reasonable cost, leading to tradeoffs. For example, a number of large cohort studies such as the English Longitudinal Study of Ageing, HRS, and others [e.g., [7]] originally chose to utilize dried blood spots, which offer a convenient and minimally invasive alternative to venipuncture, yet less precision and versatility than whole blood. Moreover, many large cohort studies continue collecting biomarkers that were most relevant in the decades they began, but now may be considered outdated. While a strength of these studies is sample sustainability and longitudinal collection of the same biomarkers over time, this limits the adoption of new technologies and the multi-modal integration of biomarkers that could improve discovery and rigor. The rise of team science and consortia The dichotomy between individual investigator-driven research, which is often highly innovative, but limited by time, money, and sample size, and large cohort studies, which have far greater statistical power and external validity, but often utilize cost-effective, less contemporary methods, has given rise to the development of a variety of exciting collaborative initiatives in recent years. These initiatives attempt to strike a balance between the need for large samples to provide statistical power and the scientific ethos of using the highest quality biological data. Indeed, in the last decade and a half, team science has flourished, particularly within the context of the study of psychiatric genetics [e.g., [8], [9], [10]]. Collaborative initiatives, such as the Psychiatric Genomics Consortium (PGC) [11], Latin American Genomics Consortium [12], The Brainstorm Consortium [13], and many others, have transformed the landscape of biobehavioral research by pooling data across studies, sometimes from around the globe, enabling unprecedented statistical power and the discovery of robust biological signals. These team science approaches have resulted in the largest genome-wide association studies (GWAS) in psychiatry to date. For instance, the Major Depression Disorder (MDD) working group of the PGC recently published a preprint of a GWAS involving nearly 5 million participants, which identified 635 genetic variants associated with MDD [14]. Findings from large-scale analyses like this have yielded key insights into the mind-body connection and have dispelled earlier assumptions that genetic risk for a specific psychiatric disorder can be traced to a single gene. Instead, GWAS have demonstrated that psychiatric conditions are highly polygenic, meaning they are driven by the cumulative influence of many genetic variants that independently exert very small effects but together contribute meaningful risk. The statistical power required to detect the aggregate influence across all of these individual small effects necessitates large sample sizes, which is made feasible by pooling data across multiple studies. Consequently, modern team science efforts emphasize collaborative approaches to understanding the complex, multifactorial nature of genetic risk. There are numerous compelling examples of this in the mood and anxiety literatures [e.g. [15], [16]]. While this consortia and team science approach has proven valuable in genetics research, the degree to which such an approach can be successful in the context of other health-relevant biomarkers depends on the ability to integrate data from many studies that use similar and, ideally, the highest quality methods across data collection, storage, processing, and analysis. In the following sections, we outline methodological considerations and best practices for incorporating biological data into behavioral research. Adoption of standardized protocols will better enable biobehavioral data to be pooled across studies and has the potential to enhance reproducibility and scientific rigor. Methodological considerations in biological data collection Robertson and colleagues [17] recently published a comprehensive, structured glossary of commonly used biomarkers in biobehavioral research, including easy-to-understand definitions, descriptions of how each biomarker is measured, and key considerations for use. We refer readers to that paper for a detailed description of specific biomarkers across a range of physiological systems (i.e., circulatory/cardiovascular, digestive, endocrine, immune, muscular and respiratory). In this section, we expand on methodological considerations in obtaining peripheral biofluids (e.g., blood, saliva, urine), which can be used to index a range of biological and health-related outcomes. Although many modalities exist for indexing biological data (e.g., postmortem tissue, neuroimaging, swabbed biospecimens, induced pluripotent stem cells), this paper focuses specifically on peripheral biofluids, which are found outside of the central nervous system. They can be obtained from living human subjects at relatively low cost using non-invasive to minimally invasive methods and without needing specialized expensive equipment. These benefits make peripheral biofluids ideal samples to incorporate into both large population analyses as well as lab-based experimental designs. In conjunction with their feasibility, peripheral biofluids can be used to index a wide range of biomarkers reflecting multiple biological pathways, including cytokines (inflammation), hormones (e.g., cortisol, estradiol), metabolites (e.g., glucose, lipids), and DNA and RNA (e.g., genetic, epigenetic, and transcriptomic analyses). This versatility makes them especially well-suited for multimodal investigations and team science initiatives that integrate biological, psychological, and behavioral data to answer a range of clinically relevant questions. Here, we offer a summary of methodological considerations and best practices for obtaining fluid peripheral biomarkers, with a particular focus on maximizing reproducibility and harmonization across studies, as each type of specimen presents different needs and challenges for storage and extraction. Specific biofluids are discussed in detail below; for a consolidated overview of common biomarkers by biofluid type, see Table S1. Blood Uses. Blood is sensitive to changes in physiology and is rich in metabolites and lipoproteins, providing biological information at a systematic level. Its composition can vary significantly over time (i.e., phasic) depending on physiology, environmental context, and biological rhythms. Thus, blood samples can provide a wide range of biomarkers, depending on what systems or analytes are under examination. For example, complete blood count (CBC; sometimes referred to as full blood count, or FBC) can provide indicators of the basic compositions of white and red blood cells, such as hemoglobin, hematocrit, monocytes, and natural killer cells, providing a snapshot of hematologic and immune health. Metabolic information can be obtained from glucose, liver enzymes, electrolytes, and creatinine (among many others). Lipid profiles (cholesterol, triglycerides) can provide indicators of cardiovascular risk. Inflammatory markers, hormones, and vitamin levels can also provide insight into several systems linked with both physical and mental health. In addition, DNA and RNA can be extracted from white blood cells for genetic and epigenetic analysis. Given that certain sampling methods, such as venipuncture, have the potential to provide a higher volume than is necessary to assess any one biomarker, a key advantage of blood is that multiple biomarkers can be assayed from one sample and leftover sample can be stored for future analysis. Collection. Blood collection methods vary based on purpose, setting, and volume of blood needed. Common methods include venipuncture (standard blood draw using a needle into a vein), capillary sampling (finger or heel stick using a microneedle/microfluid devices), and dried blood spot (spots from a fingerstick are blotted onto treated filter paper and dried). Selection of blood collection methods is accompanied by a tradeoff between analytical depth and feasibility, as each method requires specific standard procedures and protocols and unique consideration for storage and later analysis. Venipuncture is often considered the gold standard method for collecting blood, as it provides high quality, high yield samples suitable for a wide range of biomarkers; however, it is also invasive and resource-intensive, requiring a clinical infrastructure and it is typically conducted by a trained, certified phlebotomist or trained research staff member/medical professional [18]. Conversely, capillary sampling is less invasive and more feasible for field work, vulnerable populations, or repeated measures, yet yields smaller sample volumes and is more subject to variability [19]. Dried blood spotting offers the lowest participant burden and remains stable at room temperature for many analytes, simplifying storage and transport, making it ideal for large-scale or remote data collection; however, this approach offers very limited analyte range and sensitivity [20]. Components for Collection. Blood has four main components: plasma, red blood cells, white blood cells, and platelets (see Fig. 1). Plasma is the liquid component, which functions to transport blood cells and other constituents (e.g., nutrients, hormones, proteins) throughout the body, and contains clotting factors (i.e., fibrinogen). Platelets are cell fragments essential for blood clotting, binding to fibrinogen when activated to produce clots. Serum is the liquid part of blood after it has clotted and therefore does not contain fibrinogen. Blood begins to clot almost immediately following venipuncture and complete clotting usually occurs within 30 min at room temperature. Hence, the choice of blood derivative must be made early on. If blood is collected into tubes containing anticoagulants, coagulation is prevented, and the components are kept in suspension. Upon settling, the heavy red blood cells (responsible primarily for transporting oxygen throughout the body and therefore not commonly used to assay biomarkers) sink to the bottom leaving plasma – a clear liquid at the top – which contains all the components of blood except the cells. When centrifuged, a layer of white blood cells (i.e., the buffy coat) will form between the plasma and red blood cells. White buffy coat is often used to extract DNA and RNA, and to estimate white blood cell composition. When whole blood is left to clot in the absence of an anticoagulant, red blood cells can be removed as a solid mass and serum remains as a pale, yellow liquid. Allowing adequate clotting time before processing is important to ensure sample quality and accurate results.Fig. 1Blood specimen processing pathways and derived fractions for use in psychiatric research. Blood specimen processing pathways and derived fractions for use in psychiatric research. Plasma and serum are commonly analyzed in human subjects research because they contain many of the circulating biomarkers measurable in blood, including proteins, hormones, metabolites, immune signals, nutritional indicators, and markers of environmental exposure. As neither plasma nor serum typically contain enough genetic or transcriptomic data to extract (unless targeting cell-free DNA specifically), DNA and RNA are primarily extracted from white blood cells. DNA and RNA data can be extracted via the buffy coat or through whole blood lysis in anticoagulated blood or via white blood cells contained in the clot in coagulated blood. Coagulated blood tends to yield lower DNA compared to anticoagulated whole blood, though specialized kits for extracting DNA from clotted blood are often sufficient for many applications. Tubes. Blood collection tubes are designed with different additives depending on whether the priority is serum, plasma, or DNA/RNA isolation (see Figure S1). Serum tubes include plain red-top tubes, which contain no anticoagulant and allow blood to clot naturally, and serum separator tubes (SSTs, often gold-top), which contain a clot activator and gel that separates serum from cells after centrifugation. Both types of serum tubes have different advantages and disadvantages. The plain red-top tubes are useful for highly sensitive assays that may be sensitive to additives, including specific hormones (e.g., estradiol, testosterone, cortisol), certain cytokines and small proteins (e.g., IL-6, TNF-α), and fat-soluble vitamins (e.g., Vitamins A and D). Red-top tubes require longer processing time to allow for the blood to clot naturally. Alternatively, the SST tubes allow for faster clotting and easier serum separation, resulting in greater convenience and sample stability. The tradeoff is that these tubes contain slightly more additives. As such, SST tubes may be the more optimal choice for routine biochemical (e.g., glucose, electrolytes, liver enzymes, etc.) and immunochemistry (e.g., antibodies, many hormones) assays, as well as other stable inflammatory markers (e.g., CRP) and stable proteins (e.g., albumin). Plasma tubes contain anticoagulants to prevent clotting. EDTA tubes (lavender-top) are widely used for hematology and are preferred for DNA extraction, while heparin tubes (green-top) are suitable for plasma assays and flow cytometry, but are less ideal for molecular studies involving polymerase chain reaction because heparin can inhibit DNA amplification. Citrate tubes (blue-top) are used primarily in coagulation testing, as they preserve clotting factors. For nucleic acid studies, specialized stabilization tubes are essential: PAXgene DNA tubes stabilize nucleated cells for DNA analysis, PAXgene RNA tubes and Tempus RNA tubes preserve RNA integrity for transcriptomic research, and cell-free DNA tubes (e.g., Streck) prevent white blood cell lysis to allow accurate measurement of circulating DNA. Additional specialty tubes include gray-top tubes with fluoride/oxalate for glucose testing, which inhibit glycolysis, and royal blue-top tubes for trace element analyses, such as heavy metals. In all cases, the choice of tube is closely tied to downstream applications, as the wrong additive can compromise sample quality and assay results. Processing & Storage. The processing of blood samples requires the use of standardized operating procedures to ensure stability of analytes. Storage and transport conditions (e.g., the temperature and environment in which samples are held and the time until processing) can significantly impact sample degradation and therefore, accurate measurements of analytes. Cells should be separated from plasma/serum as soon as possible (ideally within 45 min of collection) to avoid ongoing metabolism, and this timing should be tracked and recorded. Blood processed for serum should be allowed to clot at room temperature for 20–30 min prior to centrifugation. Samples can be transported at a refrigerated temperature (e.g., 4°C) if they are to be analyzed fresh, or samples can be frozen (e.g., −20°C for short-term [< 1 month] or −80°C for long-term [> 1 month]) to reduce metabolic processes and to avoid continued degradation prior to analysis, though this may vary depending on the analyte of interest. In particular, DNA and DNA methylation data tend to be far more stable over long periods of time compared with RNA and other molecular analytes when stored at −80 °C, as RNA is highly susceptible to RNase-mediated degradation. Similarly, proteins and metabolites may also undergo significant enzymatic degradation or chemical modification during storage. For researchers who may have unpredictability in the length of time before centrifugation is possible, use of EDTA plasma samples is recommended, as the impact of lag time is less profound with these samples [21]. Freeze-thaw cycles should be minimized. Temporary or short-term storage conditions will vary by specimen and the integrity of analytes from collection to analyses should be validated by the laboratory to ensure stability under different conditions. Potential Confounds. The range of potential confounds in blood-based biomarkers depends on the biomarker being indexed. Potential confounds associated with blood-based biomarkers are summarized at a broad level in Table 1, with additional details provided in Table S2. Please see Figure S2 for an overview of considerations surrounding the workflow for handling and analyzing blood samples.Table 1Comparison of Peripheral Biofluids (Blood vs. Saliva vs. Urine).CategoryBloodSalivaUrinePrimary UsesHematologic assessment, immune/inflammatory processes, metabolic markers, endocrine and hormonal assessment, liver and kidney function, electrolyte balance, nutritional status, genetic/epigenetic/transcriptomic analyses, circulating proteins and growth factors, environmental exposure and toxicology, multi-omic biomarker researchEndocrine and hormonal assessment (particularly “free” hormone levels); stress-related biomarkers; genetic/epigenetic/transcriptomic analyses; some immune markers; repeated sampling designs; remote or field-based data collectionStress hormones and catecholamines; metabolic byproducts and metabolites; inflammatory markers; environmental exposure and toxicology; urinary microbiome; and cumulative biomarker measurement (e.g., 24-hour collection repeated assessments)Collection MethodVenipuncture, capillary sampling (e.g., fingerstick), dried blood spotsPassive (drooling/spitting into collection tube) or active (absorbent swab, stimulated saliva production); Assess pH or salivary flow rate when required for assay validation or flow-rate-dependent biomarkersSpot urine single sample (higher variability) or 24-hour continuous urine collection (lower variability); use of sterile container and mid-stream collection or catheterInvasivenessLow (fingerstick, dried blood spot) to moderate (venipuncture)Non-invasiveLow (sterile container, mid-stream collection) to high (catheter)Required PersonnelTrained phlebotomist or medical personnel for venipuncture; trained laboratory personnel or technician may be needed for centrifugation, aliquoting, and specimen storage. Minimal training may be sufficient for fingerstick, heel stick, or dried blood spot collectionCan require oversight from study personnel or samples can be self-collectedCan require oversight from study personnel or samples can be self-collectedPreparatoryProceduresSelect collection tube based on downstream application; standardize collection timing when relevant; prepare materials for centrifugation, aliquoting, and storageInstruct participants to avoiding food, alcohol, nicotine, and caffeine before collection; rinse mouth with water 10 min before obtaining sampleProvide instructions for spot, first morning, or 24-hour collection; use sterile procedures for microbiome or infection studiesProcessing RequirementsUnless analyzing whole blood, requires centrifugation and aliquoting to separate serum, plasma, or buffy coat; processing time should be standardized and recordedProcessing depends on target analyte and collection device; may require centrifugation, aliquoting, refrigeration, freezing, or use of stabilization buffers for DNA/RNAUsually minimal processing; may require mixing, aliquoting, creatinine or specific gravity measurement for hydration adjustmentStorage RequirementsRefrigeration at 4°C for fresh analysis; −20°C for short-term storage (<1 month) or −80°C for long-term storage (>1 month); minimize freeze–thaw cyclesStorage depends on the target biomarker (e.g., up to 72 h at room temperature for some cortisol collection kits vs. weeks to months at room temperature for some genetic collection kits); best practice remains aliquoting and storing samples at −80°C as soon as possibleRefrigeration at 4°C for short term; −80°C for long-term storageAdvantagesWide range of biomarkers; high sensitivity; multiple assays from a single sample; gold standard for many biomarkers; standardized clinical reference ranges availableNon-invasive; easy remote collection; suitable for repeated sampling; reflects biologically active and “free” hormone levelsNon-invasive; useful for 24-hour or cumulative biomarker monitoring; relatively stable for some analytes after collection when properly storedLimitationsModerately invasive; requires trained/paid personnel and clinical infrastructure; higher cost; participant burdenLower biomarker concentrations; limited analyte range compared to blood; sensitive to contamination (food, oral health, time of day); possibility of increased measurement errorAffected by hydration and diet; high variability; large volume sometimes requiredParticipant BurdenModerateLowLowCostHighLow to moderateLowBest Use CasesHigh-precision biomarker measurement; multi-omics (genomics, epigenomics, transcriptomics, proteomics); inflammation and immune markers; metabolic and cardiovascular markers; studies requiring large sample volumes or multiple assays from a single sampleField studies and remote sampling; pediatric or needle-averse populations; repeated sampling and diurnal hormone studies; stress reactivity paradigms; genetic studies using mail-in kits; and studies with limited clinical infrastructureLongitudinal monitoring; pediatric or needle-averse populations; cumulative biomarker measurement (e.g., 24-hour collection); catecholamines and dopamine; hydration-adjusted biomarkers; and studies of metabolites or environmental exposuresTemporal ResolutionSingle time-point snapshotGood for repeated and diurnal samplingGood for cumulative measurementAnalyte ConcentrationHighLowModeratePotential ConfoundsTime to processing, storage temperature, freeze-thaw cycles, hemolysis, tube type/additives, and time of dayTime of day; recent food, caffeine, alcohol, or nicotine intake; oral health or bleeding; physical activity; and cell compositionHydration level, diet, time of day, collection method, total volume, and contaminationNote. Blood provides the most comprehensive assessment of systemic biological processes but is more invasive and resource intensive. Saliva and urine provide noninvasive alternatives that are well suited for field studies, remote collection, and repeated sampling designs, though they may have lower analyte concentrations or additional confounds that should be standardized, recorded, or statistically controlled. Different biofluids reflect different physiological processes and time scales (e.g., circulating levels vs. cumulative excretion), which should be considered when selecting a biofluid. Common biofluid-specific confounds are summarized here; detailed recommended control procedures are provided in Table S1. Selection of a biofluid should be guided by the research question, target biomarkers, required temporal resolution, participant burden, and available resources. BDNF = brain-derived neurotrophic factor; CRP = C-reactive protein; DHEA = dehydroepiandrosterone; DNA = deoxyribonucleic acid; IgA = immunoglobulin A; IL = interleukin; RNA = ribonucleic acid; TNF-α = tumor necrosis factor alpha; WBC = white blood cell; °C = degrees Celsius. Table 1Comparison of Peripheral Biofluids (Blood vs. Saliva vs. Urine).CategoryBloodSalivaUrinePrimary UsesHematologic assessment, immune/inflammatory processes, metabolic markers, endocrine and hormonal assessment, liver and kidney function, electrolyte balance, nutritional status, genetic/epigenetic/transcriptomic analyses, circulating proteins and growth factors, environmental exposure and toxicology, multi-omic biomarker researchEndocrine and hormonal assessment (particularly “free” hormone levels); stress-related biomarkers; genetic/epigenetic/transcriptomic analyses; some immune markers; repeated sampling designs; remote or field-based data collectionStress hormones and catecholamines; metabolic byproducts and metabolites; inflammatory markers; environmental exposure and toxicology; urinary microbiome; and cumulative biomarker measurement (e.g., 24-hour collection repeated assessments)Collection MethodVenipuncture, capillary sampling (e.g., fingerstick), dried blood spotsPassive (drooling/spitting into collection tube) or active (absorbent swab, stimulated saliva production); Assess pH or salivary flow rate when required for assay validation or flow-rate-dependent biomarkersSpot urine single sample (higher variability) or 24-hour continuous urine collection (lower variability); use of sterile container and mid-stream collection or catheterInvasivenessLow (fingerstick, dried blood spot) to moderate (venipuncture)Non-invasiveLow (sterile container, mid-stream collection) to high (catheter)Required PersonnelTrained phlebotomist or medical personnel for venipuncture; trained laboratory personnel or technician may be needed for centrifugation, aliquoting, and specimen storage. Minimal training may be sufficient for fingerstick, heel stick, or dried blood spot collectionCan require oversight from study personnel or samples can be self-collectedCan require oversight from study personnel or samples can be self-collectedPreparatoryProceduresSelect collection tube based on downstream application; standardize collection timing when relevant; prepare materials for centrifugation, aliquoting, and storageInstruct participants to avoiding food, alcohol, nicotine, and caffeine before collection; rinse mouth with water 10 min before obtaining sampleProvide instructions for spot, first morning, or 24-hour collection; use sterile procedures for microbiome or infection studiesProcessing RequirementsUnless analyzing whole blood, requires centrifugation and aliquoting to separate serum, plasma, or buffy coat; processing time should be standardized and recordedProcessing depends on target analyte and collection device; may require centrifugation, aliquoting, refrigeration, freezing, or use of stabilization buffers for DNA/RNAUsually minimal processing; may require mixing, aliquoting, creatinine or specific gravity measurement for hydration adjustmentStorage RequirementsRefrigeration at 4°C for fresh analysis; −20°C for short-term storage (<1 month) or −80°C for long-term storage (>1 month); minimize freeze–thaw cyclesStorage depends on the target biomarker (e.g., up to 72 h at room temperature for some cortisol collection kits vs. weeks to months at room temperature for some genetic collection kits); best practice remains aliquoting and storing samples at −80°C as soon as possibleRefrigeration at 4°C for short term; −80°C for long-term storageAdvantagesWide range of biomarkers; high sensitivity; multiple assays from a single sample; gold standard for many biomarkers; standardized clinical reference ranges availableNon-invasive; easy remote collection; suitable for repeated sampling; reflects biologically active and “free” hormone levelsNon-invasive; useful for 24-hour or cumulative biomarker monitoring; relatively stable for some analytes after collection when properly storedLimitationsModerately invasive; requires trained/paid personnel and clinical infrastructure; higher cost; participant burdenLower biomarker concentrations; limited analyte range compared to blood; sensitive to contamination (food, oral health, time of day); possibility of increased measurement errorAffected by hydration and diet; high variability; large volume sometimes requiredParticipant BurdenModerateLowLowCostHighLow to moderateLowBest Use CasesHigh-precision biomarker measurement; multi-omics (genomics, epigenomics, transcriptomics, proteomics); inflammation and immune markers; metabolic and cardiovascular markers; studies requiring large sample volumes or multiple assays from a single sampleField studies and remote sampling; pediatric or needle-averse populations; repeated sampling and diurnal hormone studies; stress reactivity paradigms; genetic studies using mail-in kits; and studies with limited clinical infrastructureLongitudinal monitoring; pediatric or needle-averse populations; cumulative biomarker measurement (e.g., 24-hour collection); catecholamines and dopamine; hydration-adjusted biomarkers; and studies of metabolites or environmental exposuresTemporal ResolutionSingle time-point snapshotGood for repeated and diurnal samplingGood for cumulative measurementAnalyte ConcentrationHighLowModeratePotential ConfoundsTime to processing, storage temperature, freeze-thaw cycles, hemolysis, tube type/additives, and time of dayTime of day; recent food, caffeine, alcohol, or nicotine intake; oral health or bleeding; physical activity; and cell compositionHydration level, diet, time of day, collection method, total volume, and contaminationNote. Blood provides the most comprehensive assessment of systemic biological processes but is more invasive and resource intensive. Saliva and urine provide noninvasive alternatives that are well suited for field studies, remote collection, and repeated sampling designs, though they may have lower analyte concentrations or additional confounds that should be standardized, recorded, or statistically controlled. Different biofluids reflect different physiological processes and time scales (e.g., circulating levels vs. cumulative excretion), which should be considered when selecting a biofluid. Common biofluid-specific confounds are summarized here; detailed recommended control procedures are provided in Table S1. Selection of a biofluid should be guided by the research question, target biomarkers, required temporal resolution, participant burden, and available resources. BDNF = brain-derived neurotrophic factor; CRP = C-reactive protein; DHEA = dehydroepiandrosterone; DNA = deoxyribonucleic acid; IgA = immunoglobulin A; IL = interleukin; RNA = ribonucleic acid; TNF-α = tumor necrosis factor alpha; WBC = white blood cell; °C = degrees Celsius. Comparison of Peripheral Biofluids (Blood vs. Saliva vs. Urine). Note. Blood provides the most comprehensive assessment of systemic biological processes but is more invasive and resource intensive. Saliva and urine provide noninvasive alternatives that are well suited for field studies, remote collection, and repeated sampling designs, though they may have lower analyte concentrations or additional confounds that should be standardized, recorded, or statistically controlled. Different biofluids reflect different physiological processes and time scales (e.g., circulating levels vs. cumulative excretion), which should be considered when selecting a biofluid. Common biofluid-specific confounds are summarized here; detailed recommended control procedures are provided in Table S1. Selection of a biofluid should be guided by the research question, target biomarkers, required temporal resolution, participant burden, and available resources. BDNF = brain-derived neurotrophic factor; CRP = C-reactive protein; DHEA = dehydroepiandrosterone; DNA = deoxyribonucleic acid; IgA = immunoglobulin A; IL = interleukin; RNA = ribonucleic acid; TNF-α = tumor necrosis factor alpha; WBC = white blood cell; °C = degrees Celsius. Saliva Uses. Saliva samples are increasingly used for biomarker collection as they provide multiple advantages over blood. Many research participants prefer to provide a saliva sample over a blood sample due to the pain and anxiety sometimes associated with venipuncture, particularly among pediatric or adolescent cohorts. Similarly, researchers may prefer saliva when collecting samples in areas that do not have access to trained phlebotomists or specialized medical personnel, or in areas that do not have access to laboratory equipment and freezers for temporary storage (e.g., field studies). Saliva-based collection kits can also be sent through the mail, enabling remote collection of samples. Saliva can be used to assess hormones and stress-related markers, such as cortisol, dehydroxyepiandrosterone (DHEA), and α-amylase. A key advantage is that hormones measured in saliva are considered “free” (i.e., biologically active), whereas hormone levels assessed via blood contain both biologically active forms of the hormone and inactive forms attached to specific binding proteins. Although biomarkers of immune functioning and inflammation can be measured via saliva, they are often present at much lower concentrations compared to blood and tend to be sensitive to numerous confounds, including oral health conditions, local oral or mucosal inflammation, and recent eating or drinking behaviors [22]. As such, saliva-based immunological biomarkers often require high-sensitivity assays and are accompanied by an increased risk of measurement error compared to blood. Saliva can also be used to assess DNA and RNA, which can be used to interrogate genetics, epigenetics, telomeres, or gene expression. DNA and RNA originate from both epithelial cells lining the mouth and from leukocytes involved in immune surveillance of the mouth. Collection. Specialized kits are available for stabilization of saliva based on the intended analyte (e.g., Oragene® kits for DNA/RNA stabilization). Saliva can be collected passively by drooling or spitting into a collection tube with the help of a collection device or funnel, or actively by using absorbent swabs (e.g., Salivette®) or methods to stimulate saliva production (e.g., paraffin). Although passive methods are least subject to possible contamination, they can be time-intensive or prove difficult for some participants, whereas active methods tend to produce larger volumes more quickly. These methods may prove useful among participants that have trouble spitting, including very young children. A systematic review by Mortazavi and colleagues [23] provides a comprehensive overview of collection, transportation, preparation, and restoration techniques of human saliva. Preparation. Prior to collecting saliva, participants should be instructed to follow several procedures and asked specific questions to limit contamination of the sample and address potential confounds. First, it is recommended that participants not eat a major meal within an hour of sample collection and avoid food and nicotine intake altogether within 30 min of sample collection. Regardless of timing of their last meal, it is recommended that participants rinse out their mouths using water to remove food residue, allowing for 10 min after rinsing to minimize sample dilution. Participants should be instructed to avoid foods high in sugar, acid, or caffeine immediately prior to sample collection, as these substances may lower saliva pH and increase bacterial growth. Consumption of alcohol, caffeine, nicotine, and medications taken within the past 12 hours should be documented, as should any vigorous physical activity and the presence of oral diseases or injury. [24] Timing. Ideally, it is helpful to gather samples around the same time of day for all participants. Morning collections maximize the likelihood of equal contributions from various salivary glands; however, preferred time of day for saliva collection depends on the analyte of interest (for more details, see [23]). Storage. Kits designed to collect saliva for measuring cortisol are often stable for 72 h at room temperature, while kits designed to collect saliva for genetic analyses are often stable at room temperature for months to enable collection outside of a laboratory setting. However, it remains the best practice to aliquot these samples and store them at −80°C as soon as possible. To date, there are no established guidelines regarding the transportation and storage of human saliva samples [23]. Potential Confounds. Some biomarkers, including cortisol, fluctuate throughout the day, so, in addition to instructing participants to provide the sample at a specific time of day (typically in relation to their waking), it is important to capture the time a sample was collected so that can be considered in the study design or analysis. For epigenetic or gene expression studies using saliva, the proportion of epithelial and leukocyte cells can vary from person to person and within people over time. Thus, estimation of those proportions is also important for downstream analysis. See Figure S3 for an overview of considerations surrounding the workflow for handling and analyzing saliva samples. Urine Uses. Urine can be used to detect a wide range of peripheral biomarkers implicated in stress related disorders and chronic health conditions. Common stress and mood-related biomarkers that can be acquired from urine include cortisol, various interleukins (particularly IL-6 and IL-8), TNF-alpha, neuropeptide Y (NPY), and varying catecholamines, including epinephrine, norepinephrine, and dopamine [25]. Relative to other modalities, urine is particularly helpful for assessing levels of dopamine [26]. The urinary microbiome (urobiome) has also been investigated in certain psychiatric conditions [27], [28]. Urine offers notable strengths relative to other markers, such as being a noninvasive method of collecting samples that presents minimal discomfort, as well as convenience for collection. The noninvasive nature also makes urine a helpful method of specimen collection for children, individuals sensitive to needles, and individuals who need to monitor certain biomarkers frequently overtime. Further, when used in the context of 24-hour collection, it allows for continuous monitoring of biomarkers, allowing for a clear picture of how these markers fluctuate over time. While blood samples are often subject to the body’s tendency to seek allostasis in response to stressors, urine is a biproduct often not subjected to the body’s regulatory mechanisms, which can allow for clear and accurate assessment of these markers and how they change across time [29], [30]. Collection. Urine is typically collected as either a spot urine single sample or through 24-hour continuous monitoring. 24-hour collection typically has the lowest variability, while spot urine typically shows high variability in sample concentrations, though the least variable in single-sample urine occurs when the sample is from the first morning urine [31]. Urine can be easily collected by non-medical personnel, including the individual providing the sample or with the help of family members for those who need assistance [30]. This method also allows for increased cost-effectiveness in sample collection [29]. Urine intended for microbial investigation should be collected in a sterile container, preferably using a mid-stream collection method. However, collection via transurethral catheterization may provide a more sterile sample. Because urine is generally a low-biomass specimen, relatively large volumes (30–50 mL) are typically required to ensure sufficient yield for DNA extraction [32]. Urine can be analyzed through nuclear magnetic resonance spectroscopy, mass spectrometry, or ELISA. For urobiome analysis, microbial DNA can be extracted and sequenced. Storage. Urine is known for being particularly stable under varying storage conditions, relative to other biomarker collection methods. Still, it is recommended that both 24-hour and spot samples be refrigerated, and for longer term storage that they be frozen as quickly as possible [ideally within two hours; [31]] at −80°C. Potential Confounds. Although there is a benefit of urine not being subject to the body’s regulatory systems, thus offering a clearer picture of biomarker intensity and change, it must also be noted that this makes urine subject to confounding variables. As a waste product, dietary habits and hydration levels can both impact these samples and must be accounted for. Hydration levels can dilute or concentrate levels of a given sample. One such way of accounting for hydration levels is by accounting for creatine and “specific gravity”, which is the ratio of urine relative to distilled water and is measured through refractometry [33]. Figure S4 provides an overview of considerations surrounding the workflow for handling and analyzing urine samples. For a summary of various collection methods and tradeoffs across biofluids, see Tables S3-S6. Study design and target population considerations in biomarker selection and collection Selecting which biospecimen is best for a given research question is guided by several factors that extend beyond the biomarker of interest. These include study design (e.g., experimental vs. observational), the temporal dynamics of the biological process under investigation, characteristics of the target population, and logistical constraints. Rarely is a single biofluid optimal across all of these domains. As such, careful consideration should be given to how each domain may influence biomarker measurement and investigators should select the biospecimen that best aligns with the study objectives and ensure collection methodology is aligned with the objectives. Experimental Designs. Biospecimen selection and collection procedures vary significantly depending on the nature of the study question and, relatedly, the study design. For instance, experimental laboratory paradigms (e.g., fear conditioning, cue-reactivity, psychosocial stress induction) may seek to evaluate acute biological responses to an experimental condition and therefore often require repeated collection of a biospecimen. At a minimum, experimental designs often include baseline and post-task sampling to determine change in the biomarker of interest. For example, one study examining cortisol reactivity following the Trier Social Stress Test (TSST), a well-validated psychosocial stress paradigm [34], found that individuals without a history of depression experienced significantly greater increases in cortisol following the TSST compared to individuals with a history of depression, despite an absence of group differences at baseline [35]. These findings complement existing work showing blunted cortisol reactivity among chronically depressed individuals [36]. Many experimental studies extend beyond pre- and post-task samples by including additional delayed recovery samples to capture time course of the biological process under exploration. For example, the same group of investigators evaluated cortisol responses to the TSST across seven post-stress intervals and found that cortisol levels among individuals without a history of a psychiatric disorder appeared to spike initially but then quickly habituated in the presence repeated psychosocial stressors [37]. Conversely, individuals with depression failed to habituate across stressors. This suggests that depression may lead to lower basal cortisol due to chronic activation of the sympathetic nervous system, which, over time may lead to dysregulation of the HPA-axis negative feedback loop and therefore precludes appropriate termination of the stress response. Thus, the inclusion of repeated biomarker measurements in an experimental study design could allow for a more complete understanding of the time course associated biological processes under examination. Standardization of Baseline Sample Collection. Because changes in a biomarker can only be accurately interpreted relative to baseline levels, careful standardization of baseline sample collection is imperative. As the research environment itself may elicit physiological changes independent of the research question (e.g., anticipatory stress), many studies allow for an acclimation period (e.g., seated rest) prior to obtaining the baseline sample. Moreover, some researchers may set up their protocols such that the experimental task instructions are postponed until after the baseline sample is obtained to minimize potential physiological reactions to the task. More detailed information and recommendations related to maximizing the standardization of baseline biospecimen collection are published elsewhere [38], [39], [40]. Additional recommendations for standardization of biospecimen handling and storage are discussed below. Sampling Schedules. Various biomarkers differ considerably in their temporal kinetics, requiring sampling schedules to be tailored according to the expected kinetics of the biomarker(s) of interest and the experimental paradigm to ensure peak responses and recovery trajectories are adequately captured. For instance, although catecholamines peak within minutes and decline rapidly [41], salivary cortisol exhibits a delayed response following acute stress, typically peaking approximately 20–30 min after onset [42], and inflammatory cytokines may not peak until one to several hours later [43]. Therefore, collecting a post-task sample immediately after the experimental condition may completely miss the biological response for biomarkers with delayed temporal profiles. As such, protocols should be tailored according to the temporal kinetics of the biomarker(s) being measured and best practices would include, minimally, a baseline assessment, post-task assessment (specified according to the biomarker’s temporal characteristics), and recovery samples. Observational Designs. Unlike experimental laboratory paradigms, observational studies (e.g., ambulatory assessment, remote collection) may seek to capture basal or chronic biological processes, rather than acute responses to a specific condition. Given the less controlled nature of observational studies compared to experimental designs, it is essential that observational study protocols maximize standardization whenever possible. This must be done in tandem with efforts to minimize participant burden. Standardization across participants is particularly important given many biomarkers derived from peripheral biofluid exhibit diurnal variation and/or are influenced by a range day-to-day factors, including diet, medication use, physical exercise, menstrual cycle phase in biological females, and acute illness [38]. One means of minimizing potential bias due to confounding variables is to standardize data collection timing across participants as much as possible. Collecting samples immediately upon awakening, 30 min after awakening, or at bedtime can help control for variation that may arise from biomarkers affected by circadian rhythms [38]. Other means of facilitating valid comparisons across participants is to require fasting samples and to encourage participants to abstain from certain behaviors (e.g., nicotine and other drug use, physical activity) prior to sample collection. For behaviors or conditions that are unethical or impractical to standardize pre-analytically (e.g., medication use, menstrual cycle phase, acute illness), careful documentation of these factors can allow researchers to statistically covary for them when possible. When designing studies including remote biospecimen collection methods, it is advised that investigators ensure that study participants are well trained in protocols regarding timing, sample handling, temporary storage, and shipping. This may include providing training videos and detailed instructions with visual guides to ensure proper adherence. Emphasizing protocol adherence whenever possible through tools such as electronic reminders, timestamped collection devices, or time-stamped photographs can help ensure compliance and reduce error. Participant characteristics. Participant characteristics should also inform biospecimen selection. Although venous blood provides access to the broadest array of many biomarkers of interest in psychiatric research (e.g., genomic, epigenomic, transcriptomic, proteomic, metabolomic, and inflammatory biomarkers), the process of venipuncture may induce acute physiological responses, thereby making the collection procedure itself a confound. This may be particularly relevant in populations where stress reactivity is the primary outcome (e.g., trauma- and anxiety-related disorders). In these cases, less invasive alternatives, such as dried blood spotting, or selection of a different biofluid, such as saliva, may help reduce bias associated with the collection method itself, as well as decrease participant burden and risk. Furthermore, biochemical monitoring through the use of wearable devices such as continuous glucose monitors or electrochemical sensing devices may offer a less invasive, less burdensome, and more standardized approach to assessing peripheral biofluids [44]. Ultimately, selecting an appropriate biospecimen requires balancing scientific objectives, methodological rigor, participant burden, and logistical feasibility to maximize both data quality and study participation. Standardization challenges Standard collection and processing methodologies are critical, not only for interpretation of results across different sites, but also across time. Small differences in handling of specimens or data at any point may introduce bias that can impact sample viability and reliability of results. During study design, careful consideration should be given to sample characteristics, context of data collection, biological rhythms, and quality control thresholds that may additionally introduce bias or impact results. Careful consideration and documentation should ensure standardization of protocols, equipment, and supplies in the laboratory and in the field. For example, variables known to impact analytes from blood collection include sample processing times, temperature, freeze-thaw cycles, hemolysis, sample storage conditions, and additives in blood collection tubes [45]. Likewise, differences in data acquisition and recording length, software, processing, and analysis of data, may result in outcomes that cannot be directly compared [46]. For example, prior work has demonstrated that various DNA extraction methods impact telomere length estimation [47], illustrating how pre-analytical methodological decisions may introduce technical variability into downstream biomarker measurements. Establishment of best practices and standard operating procedures and transparent reporting of procedures can mitigate challenges and enhance reliability of biomarker research. Additionally, aside from germline genetic variation, it is critically important to recognize that most molecular analytes (including RNA transcripts, DNA methylation marks, proteins, and metabolites) are highly tissue- and cell-type specific. Consequently, measurements obtained from one biospecimen modality (e.g., saliva) may not reliably reflect the same biomarker in another tissue or fluid (e.g., blood or brain tissue), because local cellular composition, gene expression, and regulatory mechanisms can differ substantially. Nevertheless, some biomarkers can also be detected in fluids where they are not primarily produced, due to physiological transport, secretion, or clearance processes. For example, molecules generated in specific organs may enter the bloodstream, allowing for measurement in blood, whereas analytes detected in urine typically reflect renal filtration or excretion rather than local production. Even when implementation of best practices is not feasible, detailed and transparent documentation of methodological and analytical procedures can enhance clarity and comparability. Statistical and analytical considerations Many biomarkers require specialized expertise and computational infrastructure to implement “quality control” (QC) procedures to ensure the data are as accurate, reliable, and usable as possible prior to analysis. For instance, high-dimensional data, such as genomic, transcriptomic, and other -omic assays often contain thousands, sometimes millions of features (e.g., genomic variants, proteins) per sample, resulting in raw data outputs that are difficult to store and manage. These outputs, which are sometimes stored in files that are not readable by humans, require adequate computing infrastructure (e.g., high-performance computing) and the skills of a trained biostatistician or data scientist to create or follow standardized, reproducible analytic pipelines. QC by a trained biostatistician is considered a best-practice prior to analyzing high-dimensional data and often includes steps such as checking and correcting for contaminated or problematic samples (e.g., identifying duplicates, sample contamination, missing data, integrity thresholds), outliers, technical artifacts, and batch effects. “Batch effects” refers to systematic, non-biological variation between groups of samples in experimental data. They typically arise from technical factors, such as differences in instruments, reagents, or subtle variations in the experimental protocol. If not identified and addressed in the QC or the statistical analysis plan, they can lead to type I or type II errors and obscure the conclusions drawn from the experiment. Detailed reviews of this topic can be found in the following citations [48], [49]. Batch effects are well-recognized in studies that generate high-dimensional data, including those that use sequencing (bulk or single-cell) [48], [50], microarrays (for DNA methylation or gene expression) [51], [52], and mass spectrometers [53], [54]. However, batch effects have also been reported in other types of high-throughput data, such as cytokines and telomeres [54], [55], where intra- and inter-assay variability are recognized but not regularly reported. Best practices in experimental design, including balancing experimental layouts by the outcome of interest and other important covariates (e.g., sex, repeated measures over time, experimental condition) are crucial for preserving power when controlling for batch effects. Failure to balance the layout can create collinearity between the experimental effects and technical factors, making biological differences impossible to distinguish from technical factors. In other words, if batch is meaningfully related to an outcome of interest and not considered in the experimental design, controlling for batch effects can substantially reduce power to detect differences associated with that outcome. While strategies to normalize high-dimensional data are well-accepted in the field and part of most workflows, all experiments should be examined for batch effects, and data related to experimental batches and other sources of technical variation should be tracked. Approaches to investigate batch effects include hierarchical clustering and principal components analysis to visualize samples that may be more similar based on technical factors. Once identified, technical factors can be controlled through a variety of approaches including multiple normalization strategies, controlling for batch as a covariate in an analysis, or estimation of surrogate variables to capture unmeasured sources of variability [56]. As a given field progresses, standards for normalizing sources of technical variation become more rigorous, often resulting in specific packages or pipelines that are publicly available open access sources including R packages or GitHub sites such as https://github.com/topics/batch-effects. Independent of batch effects, it is also important to consider controlling for time-varying biological factors when identifying between-group differences or associations with continuous outcomes. There are different strategies for identifying covariates, including the use of Directed Acyclic Graphs, which are well accepted. However, there are often common sets of covariates that are adopted by a field of study. For example, GWAS studies regularly control for principal components indicative of genetic ancestry. Epigenome wide association studies (EWAS) from bulk tissues, including blood or saliva, regularly control for individual differences in cell composition, age, and sex. QC is a required step when analyzing many types of biomarkers and is essential for ensuring reproducibility and harmonization across studies. Diversity, trust, and ethical sampling of peripheral biomarkers Social, historical, and ethical factors also influence willingness to provide samples, particularly among populations that have been historically underrepresented or harmed by biomedical research, including racial and ethnic minority groups, Indigenous communities, sexual and gender minority populations, migrants, people living in poverty, and those in low-resource settings [57]. Historical examples, such as the non-consensual use of Henrietta Lacks' cells and the Havasupai Tribe case [58], illustrate how inadequate consent, secondary use of samples beyond the original purpose, and weak governance can erode trust. These concerns may be heightened in psychiatric research, where biological samples may be linked to sensitive information, including mental disorders, trauma exposure, substance use, ancestry, sexual orientation, or gender identity. Biomarker protocols should therefore incorporate culturally responsive engagement, transparent informed consent, and clear explanations of sample collection, storage, future use, sharing, export, planned genetic or epigenetic analyses, management of incidental findings, and participants' right to withdraw consent [59]. Consenting may require more detailed explanation of the participants’ rights, and who will and will not have access to their data, especially in vulnerable populations like forcibly displaced individuals, who may fear consequences like loss of migration status whether from declining to participate or from participating. Where appropriate, community advisory boards or local governance structures should guide decisions regarding sample collection, storage, secondary use, data sharing, and benefit sharing [60]. The mode of sample collection may also influence participation; minimally invasive approaches (e.g., saliva, urine, or dried blood spots) may reduce barriers. In sum, transparent consent processes, participant choice, robust privacy protections, and accountable governance should therefore be considered essential components of ethical and rigorous peripheral biomarker research. Recommendations and considerations for low-resourced populations The foregoing discussion and recommendations assume easily available financial and other resources for collecting, processing, and analyzing peripheral biomarker data. However, in low- and middle-income countries (LMIC) and other resource-constrained settings, effective sample collection, storage, and sharing, and equitable collaboration in team science, are often hampered by limited infrastructure (e.g., roads, laboratory facilities, mechanical ultra-low freezers, reliable power supplies), and inadequate resources (e.g., funding, personnel), requiring pre-emptive attention to alternate methods (e.g., collection on filter cards for storage and shipping at ambient temperature), as well as by regulatory, cultural, and ethical barriers. Moreover, 1% of the world population are forcibly displaced, and are often missing from biological research as studies of these populations are mostly epidemiological. The unique context for these groups – high mobility, high trauma exposure, chronic pre- and post- migration stress, issues with trust, and cultural nuances – requires not only methodological flexibility, but also longitudinal frameworks, as single time-point studies may obscure real-world changes in their mental health over time. A careful balance of resourcefulness, pragmatic adaptation of protocols, robust QC measures, and sound stewardship, can ensure the integrity, reliability, reproducibility and security of samples and data, and optimize large-scale harmonization efforts. Equitable access to, and use of, data in these settings also requires a commitment to increasing diversity and representation of LMIC population datasets in consortia [61]. Notably, the PGC has made significant strides in this area [e.g. [62], [63]] and collaborative efforts to build capacity and strengthen laboratory and bioinformatics capacity in LMIC are currently underway [64], [65], [66]. Research with low-resourced and displaced communities requires involvement of trusted community members and organizations in research design and consent processes considering literacy, legal, and political fears. Finally, researchers should provide participating communities with access to findings, fostering trust and long-term collaboration [67]. Future directions and recommendations Standardization of biomarker measurement presents unique challenges due to the independent needs of different study populations and study designs. Still, it is possible to standardize decision-making in determining which metrics and methods are most appropriate for a given study, regarding outcomes of interest, participant burden, time frames, and resources. This review focused on peripheral biomarkers obtained via blood, saliva, and urine, at the exclusion of other minimally invasive approaches, such as hair and nail samples, as well as digital and physiological biomarkers (e.g., heart rate variability, sleep patterns, actigraphy). These are useful, increasingly popular derivations of biomarkers that would similarly benefit from standardization and a set of agreed upon best practices [68], [69], particularly for contributions to digital biobanks that allow for data integration across numerous collaborators [70]. Conclusions The integration of biological data across disciplines is inevitable. In order to maximize the impact of such work, it is critical that researchers have a foundational understanding of the uses of some of the most common peripheral biomarkers assayed from living human subjects, as well as a working knowledge of standard operating procedures to guide design considerations, study procedures, analysis, and interpretation of findings. In choosing appropriate metrics and methods, it is paramount to have standardized protocols for specimen collection, storage, and transport, as well standardized analytic practices and pipelines regarding QC, covariates, and analytic approaches to increase consistency in analyses and interpretation of results. Such approaches are particularly critical for team-based consortia efforts which we believe will continue to revolutionize our understanding of the biology of a wide array of psychological phenomena. Declaration of Competing Interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: This manuscript represents the collaborative efforts of members of the Physical Health Working Group within the Psychiatric Genomics Consortium Posttraumatic Stress Disorder Working Group (PGC-PTSD). The PGC-PTSD is supported by R01MH106595 (PI: Nievergelt). Drs. Antonia Seligowski and Sinead Sinnot report financial support was provided by National Institutes of Health. Dr. Antonia Seligowski’s time is supported by NIH K23MH125920. Dr. Sinead M. Sinnott (NCCIH 2T32AT000051) received support from the NIH National Center for Complementary Integrative Health. Drs. Nathan Kimbrel and Kyle Bourassa report financial support was provided by US Department of Veterans Affairs. Dr. Nathan Kimbrel was supported by a VA Research Career Scientist (RCS) Award (#IK6BX006523) from the Biomedical Laboratory Research and Development Service of the Department of Veterans Affairs (VA) Office of Research and Development. Dr. Kyle Bourassa was supported by Award #IK2CX002694 from the Clinical Science Research and Development (CSR&D) Service of the Department of Veterans Affairs (VA) Office of Research and Development. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the VA, the U.S. government, or any other affiliated institution. Drs. Erika J. Wolf and Sage E. Hawn serve on the editorial board of the Journal of Mood and Anxiety Disorders. Given their roles on the editorial board, Dr. Wolf and Dr. Hawn had no involvement in the peer review of this article and had no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to another journal editor. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.