Applied Pharmacology Branch, U.S. Army Medical Research Institute of Chemical Defense, Aberdeen Proving Ground, Maryland 210105400
1 To whom correspondence should be addressed at Applied Pharmacology Branch, U.S. Army Medical Research Institute of Chemical Defense, Aberdeen Proving Ground, MD 210105400. Fax: 410.436.1960. E-mail: james.dillman{at}apg.amedd.army.mil.
Received May 10, 2005; accepted June 22, 2005
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Key Words: non-human primate; microarray; blood; rhesus macaque; cynomologus macaque; African green monkey.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
A variety of models have been used in these gene expression profiling studies, including rats, mice, and cultured human cells. These studies have been facilitated by commercially available oligonucleotide microarrays that are based on the sequenced genomes of rats, mice, and humans. Non-human primates (NHPs), particularly the rhesus macaque (Macaca mulatta), the cynomologus macaque (Macaca fascicularis), and the African green monkey (Chlorocebus aethiops, AGM), are also important animal models used in efforts to develop CWA medical countermeasures. However, gene expression profiling of these species is problematic given that the genomes of these species have not been completely sequenced and no commercially available oligonucleotide microarrays (genechips) exist. Given the high similarity between NHP and human genomes (e.g., 98.77% similarity between chimpanzee and human genomes, Fujiyama et al., 2002), it is reasonable to hypothesize that human genechips could be used for gene expression profiling of NHPs. Indeed, several studies that have successfully employed Affymetrix human genechips for gene expression profiling of NHPs have been published (Cáceres et al., 2003
; Chismar et al., 2002
; Enard et al., 2002
; Kayo et al., 2001
; Uddin et al., 2004
; Vahey et al., 2003
; Wang et al., 2004
;). These studies have used rhesus, chimpanzee, gorilla, or orangutan RNA, but to date no gene expression profiling studies are available that use AGM or cynomologus RNA. To develop methods for gene expression profiling of NHPs in support of efforts to develop CWA medical countermeasures, we compared the results of rhesus, cynomologus, AGM, and human samples analyzed using human genechips. We measured the quality control metrics (e.g., fluorescent intensity, gene detection, background, noise) of human genechips probed with RNA from each of these species. Intraspecies (i.e., within a species) comparisons were made to verify data reproducibility and data quality. Interspecies (i.e., cross-species) comparisons were made to determine the performance of NHP samples relative to human samples on a human genechip. We used these data to assess the practicality of using human genechips for gene expression profiling of these NHP species. Furthermore, we evaluated the feasibility of using gene expression profiling for interspecies comparison.
![]() |
MATERIALS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Human blood samples (n = 5) were collected in accordance with approved human use protocols at USAMRICD. The test subjects were all Caucasian males ranging in age from 2339 years of age at the time of the blood collection. All human test subjects were in apparent good health at the time of the blood collection.
Collection of blood tissue.
Whole blood tissue from each donor was collected using a 5-cc syringe and was immediately injected into a PAXgene Blood RNA Collection Tube (PreAnalytiX, Franklin Lakes, NJ). For NHP donors, whole blood tissue was withdrawn from the right or left saphenous vein. For human donors, whole blood tissue was withdrawn from the median cubital vein. Approximately 1.0 ml of whole blood tissue was obtained from the rhesus macaques, and 2.5 ml of whole blood tissue was obtained from all other donors. All samples were incubated in the PAXgene Blood RNA tube for 24 h at room temperature prior to extraction.
Isolation of RNA from whole blood tissue.
RNA was extracted from whole blood tissue according to the PAXgene Blood RNA Kit Handbook (April 2001), with minor modifications. Initial centrifugation time in step 1 of the handbook was increased from 10 min at 3000 x g to 15 min at 3000 x g to obtain a large enough pellet. After proteinase K treatment, the centrifugation time was increased from 3 min to 7 min to obtain a well-defined interface. The quality and amount of RNA was analyzed by UV spectrophotometry with a Nanodrop ND-1000 UV-Vis Spectrophotometer (Nanodrop Technologies, Wilmington, DE). All RNA was precipitated with 3 M sodium acetate, glycogen, and 100% ethanol and stored at 80°C.
Gene expression profiling.
Gene expression profiling was performed using Affymetrix Human Genome U133 2.0 Plus oligonucleotide microarrays, as described at http://www.affymetrix.com/support/technical/datasheets/human_datasheet.pdf (Affymetrix, Santa Clara, CA). Precipitated RNA was removed from the 80°C freezer, thawed on ice, and centrifuged for 15 min at 16,000 x g at 4°C. The supernatants were removed via pipette, and the pelleted RNA was washed with 75% ethanol and centrifuged for 10 min at 16,000 x g at 4°C. The supernatant was removed, and the pelleted RNA was washed a second time with 95% ethanol and centrifuged for 10 min at 16,000 x g at 4°C. The supernatant was removed and the RNA pellets were allowed to air dry at room temperature for approximately 15 min. Samples were reconstituted in 60 µl of RNase-free water and analyzed by UV spectrophotometry and by microcapillary electrophoresis using an Agilent Bioanalyzer (Agilent, Palo Alto, CA).
Because there was a limited supply of total RNA from the rhesus macaques and cynomologus macaques, two rounds of linear amplification were performed on all samples using the Bioarray RNA Amplification and Labeling System (Enzo Life Sciences, Farmingdale, NY). Briefly, 75500 ng of total RNA was used to generate first-strand cDNA. A T7-dt primer was used to prime reverse transcription and incorporate a T7 promoter sequence into the cDNA. RNA was eliminated by base hydrolysis followed by neutralization. A proprietary homopolymeric tail was added to the 3' end of the first-strand cDNA followed by chain termination. A site-specific primer complimentary to the homopolymeric tail was used to initiate second strand cDNA synthesis. After second strand synthesis, the purified double-stranded cDNA was used to perform in vitro transcription, resulting in approximately a 100-fold increase of copy RNA (cRNA). The cRNA was purified with RNeasy columns (Qiagen, Valencia, CA) and the concentration was determined via UV spectrophotometry. A maximum of 2000 ng of purified cRNA was used as the template in a second round of cDNA synthesis as described above. Purified double-stranded cDNA was used in an in vitro transcription labeling reaction using biotinylated UTP and CTP, resulting in a 100-fold increase of labeled target cRNA. The target cRNA generated from each sample was processed according to the manufacturer's recommendation using an Affymetrix GeneChip Instrument System (http://www.affymetrix.com/support/technical/manual/expression_manual.affx). Briefly, spiked controls were added to 15 µg of fragmented cRNA before hybridization at 45°C at 40 revolutions per minute (rpm) for 4045 h with 10 µg of cRNA (Sartor et al., 2004). Arrays were then washed and stained with streptavidin-phycoerythrin before they were scanned on the Affymetrix GeneChip Scanner. After scanning, array images were visually inspected to confirm scanner alignment and the absence of significant bubbles or scratches on the chip surface.
Data analysis.
Scanned output files from each array were obtained using Affymetrix GeneChip Operating Software (GCOS v 1.2). Raw signal intensities were normalized using either the GCOS algorithm (Affymetrix) followed by addition of a constant (c = 1) and log transformation (log2) or the robust multi-array averaging (RMA) algorithm (Irizarry et al., 2003). The normalized data were imported as a comma separated values (.csv) file into Partek Pro 6.0 (Partek, St. Louis, MO). The imported data were analyzed by principal component analysis (PCA) to determine the significant sources of variability in the data. For hierarchical clustering, Euclidian parameters were specified to calculate interpoint distances, single linkage was specified to calculate the intercluster distances, and a cophenetic correlation was calculated. Boolean analysis (Boole, 1848
) of the data was performed using Excel 2003 (Microsoft, Redmond, WA). Probe sets that exhibited an intensity with an associated p value < 0.05 for any given sample were called present. To generate the most stringent list of probe sets for intra- and interspecies comparison, we included only probe sets that were called present for each biological replicate within a given species (100% reproducibility). Once all comparisons were made, the probe set list was tabulated for each respective group comparison. The probe sets were imported into Onto-Express (Khatri et al., 2002
) as a text file to classify the molecular function and biological processes represented by the probe sets. For analysis of variance (ANOVA), animal type (human or NHP) was used as the factor. A resulting list of genes differentially expressed between human and NHP whole blood tissue was compiled (Bonferonni-corrected p < 0.05).
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
Analysis of Gene Expression Profiles: Hierarchical Clustering
The gene expression profiles for NHP and human whole blood tissue were analyzed by hierarchical clustering (Fig. 2). Hierarchical clustering is used to group similar objects together. At the start of the analysis each sample is considered a cluster. The two most similar clusters are combined and continue to combine until all objects are in the same cluster (termed the root). Hierarchical clustering produces a tree (dendogram) that shows the hierarchy of the clusters. The distance between the two members of the cluster determines its height. Groups of samples that are similar will be combined with short clusters, and tall clusters will separate dissimilar groups. The width of the clusters has no mathematical value.
|
Analysis of Gene Expression Profiles: Analysis of Variance
The gene expression profiles for NHP and human whole blood tissue were examined using an ANOVA. Animal type (human or NHP) was used as the factor in the ANOVA to detect genes differentially expressed in human whole blood tissue compared to NHP whole blood tissue. Genes significant to a Bonferroni-corrected p < 0.05 are reported in Table 1 of the Supplementary Data online.
Analysis of Gene Expression Profiles: Boolean Analysis
The gene expression profiles for NHP and human whole blood tissue were examined by means of an intraspecies Boolean analysis (Boole, 1848). Probe sets having a detection p value < 0.05 for all biological replicates within a given species (100% reproducibility) were included in the total number of probe sets detected (called "present") for that species. These results are summarized in Table 2. The Boolean analysis identified 6820 probe sets detected in the cynomologus group; 2643 probe sets detected in the AGM group; 2757 probe sets detected in the rhesus group; and 2303 probe sets detected in the human group. The results of interspecies comparisons (2-way, 3-way, and 4-way comparisons) of the probe set reproducibly detected in each intraspecies comparison are summarized in Table 2 and in a Venn diagram in Figure 3. In a 4-way interspecies comparison, 1079 probe sets were 100% reproducibly detected. These 1079 probe sets represent 1009 unique genes.
|
|
To identify the molecular functions and biological processes represented by the probe pair reproducibly detected in all species (100% reproducibility in both intra- and interspecies comparisons), we mapped these 1079 probe sets (Table 4 of the Supplementary Data online) to the Gene Ontology (The Gene Ontology Consortium, 2000). Table 5 of the Supplementary Data summarizes the molecular functions represented by this group of probe sets (p < 0.01), and Table 6 of the Supplementary Data summarizes the biological processes represented by this group of probe sets (p < 0.01).
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Although previous work has been published using human genechips to study NHPs, particularly rhesus, chimpanzee, gorilla, and orangutan RNA from various tissues (Cáceres et al., 2003; Chismar et al., 2002
; Enard et al., 2002
; Kayo et al., 2001
; Uddin et al., 2004
), an evaluation of the response of cynomologus macaque or AGM RNA on a human genechip is not available in the open literature. In this study we found that whole blood tissue RNA from each NHP tested generated reproducible data comparable to the data obtained using human RNA on a human genechip (Table 1). These results suggest that gene expression profiling of rhesus, cynomologus, or AGM can be performed reliably using human genechips. This conclusion is in agreement with previous studies that have examined the performance of rhesus RNA on human genechips (Chismar et al., 2002
; Wang et al., 2004
) and extends these observations to cynomologus and AGM RNA.
Because NHP RNA performed well on human genechips, we analyzed the actual gene expression profiles of the NHPs and humans. Analysis of the gene expression data by PCA revealed that each NHP whole blood tissue gene expression profile appears to be equally dissimilar to humans (Fig. 1). This could be due to similarities in the NHP gene expression profiles compared to humans, or to similarities in the hybridization characteristics of the NHP RNA on the genechip compared to human RNA. These results are supported by the hierarchical clustering analysis, in which the human group clusters away from the NHPs (Fig. 2). However, the intensity map representing interpoint distances between clusters suggests that the cynomologus group is more dissimilar to humans than is the rhesus or AGM group (Fig. 2A). As observed in the PCA, the intensity map, and the cluster dendogram, the cynomologus group shows the least intraspecies variability, and the human group shows the greatest intraspecies variability.
The low intraspecies variability of the cynomologus group may be a factor in the 2.5- to 3-fold greater number of probe sets observed as reproducibly detected in this group compared with the other groups (Boolean analysis, Table 2). Because a probe set was counted if it was detected (called "present") in all replicates for a species, lower intraspecies variability would tend to result in a higher number of probe sets counted. The tight clustering of the cynomologus group (observed by PCA and cluster analysis) may be due to any of a number of factors including genetic relatedness or exposure history. These observations would require further research to determine their significance, including gene expression profiling of completely naïve cynomologus macaques. Furthermore, a greater number of subjects in each group representing variations across gender, age, and ethnicity would be needed to assess more accurately the intraspecies biological variability.
In our interspecies comparison of the probe sets reproducibly detected across all replicates within a species, we identified a group of probe sets that is reproducibly detected across all species examined in our study. This group of probe sets maps to 1009 unique genes (Table 4 of the Supplementary Data online). Although the significance of this group of probe sets is not clear at present, it is interesting to speculate about potential uses for this group of probe sets. Because genes in this group of probe sets are reproducibly detected within and across the species studied, they may serve as controls useful in normalizing data collected from these different species using human genechips. Thus, these probe sets, or more likely a subset of these probe sets, have the potential to serve the purpose that "housekeeping" genes do in other types of experiments, such as Western blotting or PCR experiments. Although it is becoming clear that there are likely no universal housekeeping genes, housekeeping genes can be useful if they have been validated in a particular system (Bustin, 2002). This would require additional research looking at detection of these probe sets across an expanded population of test subjects and examining how detection of these probe sets may change after a chemical exposure. For now, this group of probe sets provides a source of potentially valuable normalization control genes useful for the future development of tools for interspecies comparisons.
One issue that is critical to consider when interpreting our data is the difference in the genomes and the mechanisms of gene expression between humans and NHPs. Although the genomes of humans and chimpanzees have been shown to be highly similar (98.77% similarity, Fujiyama et al., 2002), and presumably this is true of other NHP species, there are obviously still differences that may affect the interspecies detection of certain genes. In addition, focusing on genome similarity neglects the fact that gene expression profiling is based on mRNA expression and not on DNA sequence. A single gene does not necessarily generate a single transcript. Splicing variants are very common in the human, and humans and NHPs may use different splicing strategies in some genes. Recently, several publications have begun to address these issues of interspecies variation in gene expression and genomic sequence as it relates to the issue of analyzing NHP gene expression profiles with human genechips. Chismar and colleagues (2002)
used the U95Av2 human genechip and compared the expression patterns of humans with rhesus. They concluded that the percentage of detected genes (genes called "present") in the rhesus brain is lower than that of human brain, and that this is especially true for genes with lower signal intensity. Cáceres and colleagues (2003)
used the HG-U95Av2 human genechip to identify upregulated genes in the human cortex compared with those of the NHPs. Because sequence divergence could lead to an underestimation of expression levels in NHPs, they excluded 4572 probe sets that exhibited different hybridization behavior between two sets of samples in order to reduce false positives. However, this analysis is based solely on probe set signal intensities and not on actual sequence data. Wang and colleagues (2004)
employed a sequence analysis approach to assess the utility of human genechips for the study of NHP gene expression profiles. They identified probe sets conserved between rhesus and human based on sequence analysis and identified these probe sets as providing a more accurate reflection of gene expression profiles. They found that, of the 54,675 probe sets on the HG-U133 Plus 2.0 genechip (representing the entire human genome), 3636 were interspecies conserved between humans and rhesus (6.6%). Pairwise correlation coefficients of 20 samples (12 human and 8 rhesus) were calculated for expressed probe sets (0.65 ± 0.044) and for the ISC probe sets (0.80 ± 0.026). These results suggest that the reproducibility of interspecies comparisons can be increased by using a subset of probe sets that have been previously defined based on sequence analysis. This worked well for rhesus because there is a considerable amount of sequence information available in public databases. However, in the case of cynomologus macaques and African green monkeys, there is little sequence information publically available. A GenBank search on June 20, 2005, revealed 3322 total entries and 181 mRNAs deposited for African green monkeys, and 4481 total entries and 3559 mRNAs deposited for cynomologus macaques. In contrast, there were 58,815 total entries and 48,152 mRNAs deposited for rhesus macaques, and 9,279,889 total entries and 6,406,834 mRNAs deposited for humans. The disparity in sequence information among the NHPs is due to current efforts to sequence the rhesus genome, for which a draft assembly is in progress (www.genome.gov). Recent comparisons looking at specific gene families have shown that there is 5% sequence divergence between rhesus and human sequences (Gilad et al., 2003
; Wall et al., 2003
). Thus, there is still a considerable amount of work to be done in developing tools to compare the gene expression profiles of humans and NHPs.
In conclusion, we have shown that gene expression profiling of NHP samples using human genechips gives reliable, reproducible data. Comparison of humans with NHPs will become more robust as larger data sets are studied and new tools are advanced to address this challenge. This preliminary data set serves as the foundation for the genomic assessment of NHP responses to CWA exposure and medical countermeasures, and it will enhance efforts to develop CWA medical countermeasures that are safe and effective in humans.
![]() |
SUPPLEMENTARY DATA |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Supplementary Table 1. Genes differentially expressed in human whole blood tissue compared with NHP whole blood tissue; Supplementary Table 2. Molecular functions represented by each species probe pair set; Supplementary Table 3. Biological processes represented by each species probe pair set; Supplementary Table 4: Probe pair set reproducibly detected in intra- and interspecies comparisons by Boolean analysis; Supplementary Table 5. Molecular functions represented by the probe pair set reproducibly detected in intra- and interspecies comparisons; Supplementary Table 6: Biological processes represented by the probe pair set reproducibly detected in intra- and interspecies comparisons.
![]() |
DISCLAIMER |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
NOTES |
---|
The authors certify that all research involving human subjects was done under full compliance with all government policies and the Helsinki Declaration.
![]() |
ACKNOWLEDGMENTS |
---|
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Blanton, J. L., D'Ambrozio, J. A., Sistrunk, J. E., and Midboe, E. G. (2004). Global changes in the expression patterns of RNA isolated from the hippocampus and cortex of VX exposed mice. J. Biochem. Mol. Toxicol. 18, 115123.[CrossRef][ISI][Medline]
Boole, G. (1848). The Calculus of Logic. Cambridge Dublin Math. J. 3, 183198.
Bustin, S. A. (2002). Quantification of mRNA using real-time reverse transcription PCR (RT-PCR): Trends and problems. J. Mol. Endocrinol. 29, 2339.
Cáceres, M., Lachuer, J., Zapala, M. A., Redmond, J. C., Kudo, L., Geschwind, D. H., Lockhart, D. J., Preuss, T. M., and Barlow, C. (2003). Elevated gene expression levels distinguish human from non-human primate brains. Proc. Natl. Acad. Sci. U.S.A. 100, 1303013035.
Chismar, J. D., Mondala, T., Fox, H. S., Roberts, E., Langford, D., Masliah, E., Salomon, D. R., and Head, S. R. (2002). Analysis of result variability from high-density oligonucleotide arrays comparing same-species and cross-species hybridizations. Biotechniques 33, 516524.[ISI][Medline]
Dillman, J. F. III., Phillips, C. S., Dorsch, L. M., Croxton, M. D., Hege, A. I., Sylvester, A. J., Moran, T. S., and Sciuto, A.M. (2005). Genomic analysis of rodent pulmonary tissue following bis-(2-chloroethyl) sulfide exposure. Chem. Res. Toxicol. 18, 2834.[CrossRef][ISI][Medline]
Dillman, J. F. III., Dorsch, L. M., Hege, A. I., Phillips, C. S., Kiser, R. C., Choi, Y. W., and Sabourin, C. L. (2004). Genomic analysis of the mechanism of action of potential vesicant countermeasures. Toxicol. Sci. 78(S-1), 1890 (Abstract).
Draghici, S., Chen, D., and Reifman, J. (2004). Applications and challenges of DNA microarray technology in military medical research. Mil. Med. 169, 654659.[Medline]
Enard, W., Khaitovich, P., Klose, J., Zollner, S., Heissig, F., Giavalisco, P., Nieselt-Struwe, K., Muchmore, E., Varki, A., David, R., et al. (2002). Intra- and interspecific variation in primate gene expression patterns. Science 296, 340343.
Fujiyama, A., Watanabe, H., Toyoda, A., Taylor, T. D., Itoh, T., Tsai, S. F., Park, H. S., Yaspom M. L., Lehrach, H., Chen, Z., et al. (2002). Construction and analysis of a human-chimpanzee comparative clone map. Science 295, 131134.
Gilad, Y., Man, O., Paabo, S., and Lancet, D. (2003). Human specific loss of olfactory receptor genes. Proc. Natl. Acad. Sci. U.S.A. 100, 33243327.
Hotelling, H. (1933). Analysis of a complex of statistical variables into principal components. J. Educ. Psychol. 24, 498520.
Irizarry, R. A., Hobbs, B., Collin, F., Beazer-Barclay, Y. D., Antonellis, K. J., Scherf, U., and Speed, T. P. (2003). Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4, 249264.
Jolliffe, I. T. (1986). Principal Component Analysis, Springer-Verlag, New York.
Kayo, R., Allison, D. B., Weindruch, R., and Prolla, T. A. (2001). Influences of aging and caloric restriction on the transcriptional profile of skeletal muscle from rhesus monkeys. Proc. Natl. Acad. Sci. U.S.A. 98, 50935098.
Khatri, P., Draghici, S., Ostermeier, G. C., and Krawetz, S. A. (2002). Profiling gene expression using onto-express. Genomics 79, 26670.[CrossRef][ISI][Medline]
Sabourin, C. L., Rogers, J. V., Choi, Y. W., Kiser, R. C., Casillas, R. P., Babin, M. C., and Schlager, J. J. (2004). Time- and dose-dependent analysis of gene expression using microarrays in sulfur mustardexposed mice. J. Biochem. Mol. Toxicol. 18, 300312.[CrossRef][ISI][Medline]
Sartor, M., Schwanekamp, J., Halbleib, D., Mohamed, I., Karyala, S., Medvedovic, M., Tomlinson, C. R. (2004). Microarray results improve significantly as hybridization approaches equilibrium. Biotechniques 36, 790796.[ISI][Medline]
The Gene Ontology Consortium. (2000). Gene ontology: tool for the unification of biology. Nat. Genet. 25, 2529.[CrossRef][ISI][Medline]
Uddin, M., Wildman, D. E., Liu, G., Xu, W., Johnson, R. M., Hof, P. R., Kapatos, G., Grossman, L. I., and Goodman, M. (2004). Sister grouping of chimpanzees and humans as revealed by genome-wide phylogenetic analysis of brain gene expression profiles. Proc. Natl. Acad. Sci. U.S.A. 101, 29572962.
Vahey, M., Nau, M., Taubman, M., Yalley-Ogunro, J., Silvera, P., Lewis, M. (2003). Pattern of gene expression in peripheral blood mononuclear cells of rhesus macaques infected with SIVmac251 and exhibiting differential rates of disease progression. AIDS Res. Hum. Retroviruses 19, 369387.[CrossRef][ISI][Medline]
Wall, J. D., Frisse, L. A., Hudson, R. R., and Di Rienzo, A. (2003). Comparative linkage-disequilibrium analysis of the beta-globin hotspot in primates. Am. J. Hum. Genet. 73, 13301340.[CrossRef][ISI][Medline]
Wang, Z., Lewis, M. G., Nau, M. E., Arnold, A., and Vahey, M. T. (2004). Identification and utilization of inter-species conserved (ISC) probe sets on Affymetrix human GeneChip platforms for the optimization of the assessment of expression patterns in non human primate (NHP) samples. BMC Bioinformatics 5, 165.[CrossRef][Medline]
|