|Disclaimer: This is not a peer-reviewed research article but an example of laboratory report based from a biology course at California State University San Marcos.|
Authors: Benison Zerrudo, Esperanza Valle (California State University San Marcos)
Stromatolites offer an aspect of evolution which gave researchers an interest in its microbial composition. Photosynthetic bacteria are generally responsible for the formation of stromatolites in shallow waters, but there have been discoveries about the existence of stromatolites in the desert north of Mexico. Sample of stromatolites found in the Anza Borrego desert located east of San Diego was collected and analyzed resulting in morphological similarity with the marine photosynthetic bacteria Lyngbya aestuarii. Researchers deemed the morphological findings unusual because marine microorganisms such as L. aestuarii are unlikely to thrive in a desert habitat. We sought to investigate the DNA of the Anza Borrego isolates using comparative genomics. DNA isolation, DNA library generation and quality check, next-generation sequencing, genome assembly, and bioinformatics analysis were performed to procure our results. DNA sequence alignment of the isolates did not show similarities with L. aestuarii, but the isolates appeared to be genetically related to 3 species in the family of Chitinophagaceae namely Panacibacter, Lacibacter., and Filimonas. Our result may reinforce the idea that bacterial composition in stromatolites is not limited to cyanobacteria, but genome assembly quality assessment suggests possible contamination of Chitinophagaceae.
Stromatolites are layered sedimentary structures formed by photosynthetic bacteria over long periods of time in shallow waters. These bacteria perform activities that trap, bind, and/or precipitate sediments to form microbial mats. Layer by layer, these microbial mats build up and grow slowly over time (Winner, 2013). Living stromatolites are rare and can be found in a limited number of geographical locations which includes Shark Bay in Australia, Exuma Sound in Bahamas, and Cuatro Cienegas Basin in Mexico. In contrast, fossilized stromatolites are abundant and provide information of life during ancient times. Extant stromatolites can be found mostly in high-salt content lakes and marine lagoons, and they are observed as both domal-type and beds. Inland stromatolites can also be found in salt waters and waters with high ionic concentrations and poor nutrient content.
Modern stromatolites are important due to its implication on our understanding of the ancient environment of the past geological age and the origin of life. Stromatolites contribute to all the new discoveries and findings on the evolution of earth which is why they are seen as importnant organisms. They are the dominant fossil between the time during which oxygen first appeared and just before the spread of complex life such as trilobites and corals. Studies suggest that the dramatic drop of stromatolite abundance is due to the evolution of grazing multicellular organisms (Bush Heritage Australia). Stromatilaties the earliest fossils, not only are known to be the first organism that has produce free oxygen but they are valued more because they have shaped our earth in forms of the features, characteristics found on them initiating new findings bases on earth’s evolution.
Past research reported algal stromatolites from perennial streams, ponds, and rivers in northern Mexico and discovered modern stromatolites in Carrizo Gorge of Anza-Borrego Desert just east of San Diego (Buchheim, 1995). The Anza-Borrego stromatolites were only two centimeters thick, but form a large structure covering the top of granite cobbles and boulders. Cyanobacteria were isolated from the Anza-Borrego stromatolites and its morphology were analyzed. Structural features of the bacteria resemble similarity with marine photosynthetic bacteria Lyngbya aestuarii.
L. aestuarii is a filamentous, non-heterocystous photosynthetic bacteria which contributes to the intertidal microbial mats of the marine environment around the world (Kothari et. al, 2013). Researchers found that L. aestuarii is unusual due to its powerful ability to produce hydrogen. One study suggests that the genome of a strain reveals L. aestuarii can adapt to an environment with exposure to extreme solar radiation and the absence of humidity. Since L. aestuarii grows in the intertidal depths of the marine environment (M.D. & Guiry, 2020), it is unusual for researchers to discover L. aestuarii in Anza-Borrego Desert. Using genetic techniques and analysis, we aim to uncover gene sequences of the Anza-Borrego isolates and determine if the cyanobacteria is indeed L. aestuarii species. The outcome of the experiment should contribute to our understanding of the habitat and distribution of L. aestuarii species as well as the microbial components of extant stromatolites.
MATERIALS AND METHODS
CTAB Isolation of Genomic DNA
One liter of stromatolite bacterial sample (CHAM01) grown in BG11 media to late log or early stationary phase. Bacterial cells harvested using centrifugation with settings at 12,000xg for 10 minutes. Cells were ground in liquid nitrogen to break up the filaments and cellular mats. Pellet resuspended in 475μl of TE buffer. To break cell membrane and digest proteins, 25μl of 10% SDS and 2.5μl of 20 mg/ml proteinase K were added. Sample mixed and incubated at 37°C for 45 minutes. To denature protein and separate DNA, 90μl of 5 M NaCl was added and mixed. A solution of 75μl CTAB/NaCl (10% CTAB, 0.7 M NaCl) was added to the sample to separate DNA. Sample incubated for 20 minutes at 65°C. To remove proteins, the sample was cooled to room. temperature before extracting with about 675μl of chloroform/isoamyl alcohol (24:1). Sample centrifuged for 10 minutes at 12,000 RPM to separate the phases. Upper aqueous phase removed and transferred into a clean microfuge tube. Chloroform extraction repeated twice. Upper aqueous phase removed, and the nucleic acid was precipitated with the addition of 2/3 volume (about 440μl) cold isopropanol. Sample mixed and left at -70°C for 30 minutes. Nucleic acid pelleted by centrifugation with settings at 12,000 RPM for 10 minutes. Pellet washed by adding 70% ethanol and centrifuging briefly to remove leftover supernatant bound to the DNA pellet. Second 70% ethanol wash performed before allowing the pellet to air dry for about 5 to 10 minutes. Pellet resuspended with the addition of 50μl of TE buffer.
Determining the Quality of the Genomic DNA
Nanodrop spectrophotometer was used to measure the concentration of the genomic DNA (gDNA) by detecting the absorbance at wavelengths 260 nm and 280 nm. Concentration of gDNA determined using the extinction coefficient whereby A260 = 1 is equivalent to 50μl/mg of DNA. Usually, proteins absorb light at 280 nm while nucleic acids absorb light at 260 nm. Integrity of the genomic DNA (gDNA) determined by loading 200 ng of the gDNA onto an E-gel NGS (0.8% Agarose). gDNA sample on the E-gel diluted using deionized water to bring the total volume to 20μl. Marker lane created using E-Gel™ 1 Kb Plus DNA Ladder. Gel electrophoresis ran according to the manufacturer’s instructions (Thermofisher) using E-Gel NGS with the iBase™ Power System.
Fragmenting DNA is required because sequencing can only be performed for short DNA strands. In 0.2 ml PCR tube, 250 ng of genomic gDNA was pipetted. Total volume brought to 19.5μl using a low EDTA TE buffer to suspend DNA samples. Enzymatic Prep Master Mix prepared using Buffer K1 (3μl), Reagent K2 (1.5μl), and Enzyme K3 (6μl). Sample gDNA added with 10.5μl of pre-mixed Master Mix and thermocycled (hold at 4°C, 32°C for 10 min, 65°C for 30 min, hold at 4°C).
Next generation sequencing requires adaptor-ligated DNA fragments. One μl Reagent W4 diluted using 29μl low EDTA TE. Ligation Master Mix prepared using low EDTA TE (9μl), Buffer W1 (12μl), Enzyme W3 (4μl), Adaptor 1 50X (2.5μl), and Adaptor 2 70X (2.5μl). Enzymatic prep sample tube added with 30μl of the Ligation Master Mix. Sample placed in a thermocycler (20 minutes at 20°C with lid heating set to 40°C) to ligate adaptors into the DNA fragments.
MagBead Cleanup of Ligation Reaction
Magbead cleanup performed to recover adaptor-ligated DNA fragments with specific length. gDNA sample added with 48μl of Magbeads and mixed by vortexing. gDNA sample incubated for 5 minutes at room temperature. Beads pelletize using a magnetic rack for about 2 minutes. Supernatant removed and the remaining sample was added with 180μl of 80% ethanol. Sample incubated for 30 seconds and the ethanol was removed. Ethanol wash performed twice. Residual ethanol removed using pipette. Beads resuspended using a 22μl low EDTA TE buffer. Magbeads pelletized using a magnetic rack for 2 minutes. About 20μl of the supernatant containing the DNA fragments transferred to a clean tube and stored at -20°C.
Indexing PCR Master Mix created using a low EDTA TE buffer (10μl), Reagent R2 (4μl), Buffer R3 (10μl), and Enzyme R4 (1μl). Entire eluted sample and primer added with the 25μl PCR Master Mix. Sample ran in the indexing PCR thermocycler with the following settings: heated lid 105°C, 98°C for 30 sec, 98°C for 10 sec, 60°C for 30 sec, 68°C for 60 sec, hold at 4°C. Index primers were added into the DNA fragments to allow efficient DNA sequencing and sample origin identification.
MagBead Cleanup 1 of Indexing PCR Reaction
Magbead cleanup performed to recover indexed DNA fragments. Sample added with 32.5μl Magbeads and incubated for 5 minutes at room temperature. Magbeads pelletized using a magnetic rack for about 2 minutes. Supernatant removed and the remaining sample was added with 180μl 80% ethanol. Sample incubated for 30 seconds and ethanol was removed. Ethanol wash performed twice. Residual ethanol removed using pipette. Magbeads resuspended in a 22μl low EDTA TE buffer. Magbeads pelletized using a magnetic rack for 2 minutes. About 20μl supernatant transferred to the clean tube with the final library.
Magbead Cleanup 2 of Indexing PCR Reaction – Removal of Residual Primers
Magbead cleanup performed to remove residual index primers. Sample added with 24μl Magbeads and incubated for 5 minutes at room temperature. Magbeads pelletized using a magnetic rack for 2 minutes. Supernatant removed and the remaining sample was added with 180μl 80% ethanol. Sample incubated for 30 seconds and ethanol was removed. Ethanol wash performed twice. Residual ethanol removed using pipette. Magbeads resuspended using 22μl of low EDTA TE buffer. Magbeads pelletized using a magnetic rack. About 20μl of supernatant transferred to the clean test tube with the final library.
Real Time PCR Quantitation
Amplification of DNA fragments has been monitored using real time PCR. Concentration of gDNA determined using NEBNext Library Quant Kit Quick Protocol (E7630). Master Mix and Primer Mix prepared using 500μl NEBNext Library Quant Primer Mixer and 1.5ml NEBNext Library Quant Master Mix. NEBNext Library Quant Dilution Buffer (1X) prepared using 1:10 dilution of the 10X buffer in nuclease-free water. Initial 1:1000 dilution prepared for each library sample in NEBNext Library Quant Dilution Buffer (1X). One μl library sample added to 999μl NEBNext Library Quant Dilution Buffer (1X) to create 1:1000 dilution. Ten μl of 1:1000 dilution added to 90μl NEBNext Library Quant Dilution Buffer (1X) to create 1:10000 dilution. Ten μl of 1:10000 added to 90μl NEBNext Library Quant Dilution Buffer (1X) to create 1:100000 dilution. DNA standards prepared using NEBNext Library Quant Master Mix with primers (16μl) and Library Dilution Buffer 1X (4μl). Library Dilution Reactions prepared using 2X SYBR Green Master Mix (10μl), Primer 2.1 2μM (2.5μl), Primer 1.1 2μM (2.5μl), library dilution (4μl), and deionized water (1μl). No-template control prepared using 2X SYBR Green Master Mix (10μl), Primer 2.1 2μM (2.5μl), Primer 1.1 2μM (2.5μl), and deionized water (1μl). Samples and standards loaded to qPCR in 19μl volume. QPCR cycling condition set to initial denaturation (95°C, 1min, 1 cycle), denaturation (95°C, 15sec, 35 cycles), extension (63°C, 45sec).
Library Size Estimation
Library fragment size determined using E-Gel (2% Agarose). Library sample (100 ng) combined with sterile water to bring volume to 20 ul. Diluted sample loaded to E-Gel with marker lane using E-Gel™ 1 Kb Plus DNA Ladder. Electrophoresis ran with iBase™ Power System with program setting RUN E-Gel 0.8%-2.0%, default run time-26 minutes, and maximal run time-40 minutes.
Galaxy Data Analysis and Bioinformatics
Two FastQ files, 02idLYLIBRARY_S2_R1_001.fastq (Read 1) and 02idLYLIBRARY_S2_R2_001.fastq (Read 2) obtained following the gDNA library sequencing. Both files are imported to Galaxy (usegalaxy.org). Read quality for both files analyzed using FastQC with settings at default. Contigs assembled using Shovill tool with settings input read-paired end, forward reads-Read 2 file, reverse reads-Read 1 file, trim reads-off, assembler to use-velvet, list of kmer size to use-41, 43, 45, 47, 49, 51, 53, 55, 57, 59, 61, other settings-default. Genome assembly quality analyzed using Quast with lower threshold for a contig length at 1000, all other settings at default. Lyngbya aestuarii genome downloaded from NCBI Genome and uploaded to Galaxy. Control sequence alignment processed using two identical files of L. aestuarii genome with the Nucmer tool (default settings). Experimental sequence alignment processed using query sequence (Shovill contig files) and reference sequence (L. aestuarii genome) with Nucmer tool (default settings). Secondary alignment analysis performed using DNADiff (default settings). All sequences of the first contig from Shovill data file searched for similarities in NCBI Blast.
Quality and Integrity of the Isolated DNA
We determined the concentration of the isolated gDNA using a nanodrop spectrophotometer by measuring the absorbance at 260 nm (A260) and 280 nm (A280) and calculating the A260/A280 ratio. The measurements at A260 and A280 were 3.397 abs and 1.834 abs respectively, giving a A260/A280 ratio of 1.85. A 260/280 ratio of about 1.8 is generally accepted as pure for DNA. The gDNA concentration was calculated to be at 169.9 ng/µL. To determine the integrity, 200 ng of the gDNA (about 1.18 µL) was diluted and brought the total sample volume to 20 µL with sterile water, and the gDNA sample was ran through E-Gel® NGS (0.8% Agarose) using the iBase™ Power System according to the manufacturer’s (ThermoFisher) instructions. Visual inspection was performed, and we determined the presence of gDNA fragments just above the 15000 bp mark. Lane 6, 8, 10 showed a dim band of the gDNA and lane 2, 4, 12 showed a medium-to-bright band of the gDNA (Figure 1). Lane 4 and 12 exhibited shearing bands between the 250 and 2000 bp mark. The gDNA library sample of lane 8 was used in subsequent processes and analyses.
Genomic DNA Library Size Estimation and Quality Control
The three gDNA library dilutions (1:1000, 1:10000, 1:100000) and the DNA standards were prepared for quantitative PCR (qPCR). The qPCR assay was performed in a real-time thermal cycler using FAM/SYBR setting. The qPCR amplification plot (Figure 2A) and result summary were produced to determine the cycle thresholds (Ct) for all the gDNA library dilutions. The mean Ct of the 1:1000, 1:10000, and 1:100000 gDNA library dilutions were 34.658, 31.777, and 32.257, respectively. We used the melting curve analysis to determine the average dissociation temperature of the double-stranded DNA during heating. The mean dissociation temperatures of the 1:1000, 1:10000, and 1:100000 gDNA library dilutions broadly peak at 78.924°C, 75.185°C, and 80.221°C respectively (Figure 2B) suggesting slight differences in DNA fragment lengths. The mean Ct of the 4 standard samples (10 pM, 1pM, 0.1 pM, and 0.01 pM) were plotted to construct a regression line equation to determine the concentrations of the gDNA library dilutions. Using the regression line equation, we calculated the log10 concentrations of the 1:1000, 1:10000, and 1:100000 gDNA library dilutions to be -6.09, -5.29, and -5.42, respectively (Figure 2C). The antilog concentrations were calculated and multiplied by the dilution factor revealing the initial concentrations of the three dilutions to be 0.000816, 0.0516, and 0.38 (1:1000, 1:10000, 1:100000). The average concentration is 0.144 and the coefficient of variance percent is 142.92 percent. The high percent value of the coefficient of variance suggests high variation of concentrations across all dilutions which can be considered unreliable. Using qPCR Library Quantification webtool, we determined the PCR efficiency of the DNA standards to be at 89.57%, and we also determined that all mean Ct of the 1:1000, 1:10000, and 1:100000 gDNA library dilutions fall outside the standard curve range of the DNA standards. To determine the size of the gDNA library fragments, we diluted 100 ng of the gDNA library sample with sterile water, bringing the sample volume to 20 µL, and ran the sample in E-Gel® (2% Agarose) using the iBase™ Power System. Lane 3 and 4 exhibited presence of smeared gDNA fragments around the 300 bp mark. Lane 1, 3, and 6 exhibited minimal presence of gDNA fragments, while lane 2 exhibited a thick band just below the 300 bp mark (Figure 2D).
Post-Sequencing Quality Control using FastQC
FastQ files, 02idLYLIBRARY_S2_R1_001.fastq (Read 1) and 02idLYLIBRARY_S2_R2_001.fastq (Read 2) were obtained following the gDNA library sequencing. The read quality for Read 1 and Read 2 was assessed using the FastQC tool (Galaxy Version 0.72+galaxy1) in Galaxy. In per base sequence quality (Sanger/Illumina 1.9 encoding), the mean quality score for all the read positions (both in Read 1 and Read 2) is >30 (Figure 3). A quality score of 30 equates to Illumina Q30 with an error probability of 0.001. Using the FastQC raw data, we calculated the percentage of sequences with a quality score of >30 to be 95% and 88% for Read 1 and Read 2, respectively. In per-sequence GC content, both Read 1 and Read 2 have mean GC content peaks at approximately 43% and 63% which manifest a bimodal distribution on the chart (Figure 4).
Genome Assembly Quality Control
Contig files were created using the Shovill tool (Galaxy Version 1.1.0+galaxy0) in Galaxy. Post-genome assembly quality was assessed using Quast (Galaxy Version 5.0.2+galaxy1). According to the Quast data summary, the number of contigs with base pair >= 1000 is 1376, total length with base pair >= 1000 is 6738667, percent GC content is 49.76, N50 is 10030, L50 is 139, and the largest contig has a length of 119550. N50 and the largest contig are depicted on the Nx chart (Figure 5).
DNA Sequence Alignment Analysis
We performed two DNA sequence alignment using Nucmer (Galaxy Version 4.0.0beta2+galaxy0) and DNADiff (Galaxy Version 4.0.0beta2+galaxy0) to compare our query sequence to the L. aestuarii sequence obtained from NCBI Genome database. A control alignment analysis was performed using two L. aestuarii sequences which depicts a 45° angle line on the Nucmer plot signifying uniformity. The experimental alignment analysis was performed between our query sequence and the L. aestuarii sequence and the result did not depict a 45° angle line in the Nucmer plot (Figure 6). Data report was obtained from the DNADiff analysis and a control alignment analysis was performed using two L. aestuarii sequences. The experimental alignment analysis was performed using DNADiff comparing L. aestuarii and our query sequence, and the result shows only 0.13% aligned sequences on our query and only 1.60% aligned sequence on the L. aestuarii. In terms of aligned bases, our query sequence reported only 0.06% alignment while L. aestuarii sequence reported only 0.07% alignment (Table 1).
BLAST Comparison of Contig
NCBI Blast was used to find similarities of all the sequences of the first contig produced by the Shovill tool. The result shows the top 4 significant alignments (E value of 0) namely Mus musculus (90.57% identity), Panacibacter ginsenosidivorans (76.67% identity), Lacibacter spp. (76.13% identity), and Filimonas lacunae (73.34% identity). No species appeared from the Lyngbya genus within the first 100 significant alignments.
Although preceding morphological analysis concluded similarities between the isolated cyanobacteria and L. aestuarii, our findings refuted genetic relationship. The purity of the isolated gDNA was assessed by determining the A260/A280 with a value of 1.85. A ratio of about 1.8 is generally accepted as “pure” DNA (Thermofisher). In DNA integrity, we have detected the isolated gDNA fragments with different concentrations above the 15000 mark, a good indicator of preferred gDNA fragment length. Three library dilutions were created and amplified resulting in an amplification plot displaying different cycle thresholds for each dilution. The differential cycle thresholds suggest the serial dilution process was performed so that each gDNA library had different initial concentration. The calculated coefficient of variance of the concentrations was high suggesting unreliability. All the log10 concentration of the sample dilution fell outside the standard curve (all < -2) which suggests low concentration of the genomic library. Melting curve analysis was performed and the result indicates a single broad peak around 79°C for all the three library dilutions suggesting dissociation of pure double-stranded DNA with no primer dimer and/or adapter dimer contamination. Post-qPCR calculation of the initial concentration of each library dilution (1:1000, 1:10000, 1:100000) indicates varying results. The initial concentration variation of the library dilutions suggests discrepancy during the serial dilution process. The gDNA fragment length was determined pre-NGS and gel electrophoresis exhibited fragments around the 300 bp mark, the preferred DNA fragment length. Post-sequencing quality control by means of per-sequence GC content manifests a bimodal distribution which suggests two types of fragments from two different organisms. Mean quality score for all read positions is >30 which corresponds to a very low error probability. Nucmer data plot did not indicate a 45° angle line which strongly suggests correlation between our query sequence and L. aestuarii is absent. DNADiff reported a significantly low alignment percentage on both the sequences and bases indicating genetic relationship between our query sequence and L. aestuarii to be almost non-existent. NCBI Blast comparison revealed the query sequence is closely related to Mus musculus, Panacibacter ginsenosidivorans, Lacibacter spp., and Filimonas lacunae. This finding is very unusual because significant genetic relationships can easily be expected between species with similar morphology. With the aligned sequences and bases at significantly low percentages, we could be dealing with a different organism such as the species predicted on the Blast comparison result and that it reinforces the notion that microbial composition of stromatolites is not limited to cyanobacteria. The implication of the finding could help towards our understanding of the bacterial composition of modern stromatolites. It is also important not to disregard the possibility of contamination as suggested on the per-sequence GC content plot with bimodal distribution. Before we can make stable conclusions, we need to perform future experiments such that certain quality control analyses are in-compliance including high DNA integrity, unimodal distribution of GC content, and amplification of target DNA with at least 90% efficiency. By blasting only the first contig, we can conclude that the query sequence is not related to L. aestuarii but possibly related to P. ginsenosidivorans, Lacibacter spp., and F. lacunae. Making an inference of the Anza Borrego isolates being an L. aestuarii is impossible due some non-acquiescent results. Processes before and during real time QPCR should be performed properly to obtain a good pre-NGS concentration of the genomic library and that the log10 concentration of the sample falls within the standard curve range. As per sequence GC content analysis, bimodal distribution was traced suggesting possible contamination which can be mitigated through proper labeling and handling of the samples. Mislabeling of the initial bacterial culture is a possibility and that the sample provided is not in fact of cyanobacteria. Lastly, insufficient biological sequences were used to blast for comparison with other species making unsubstantial significant alignment results. Suggestion for better and reliable alignment results includes blasting an extensive proportion of the genome or whole-genome blasting.
|Disclaimer: This is not a peer-reviewed research article but an example of laboratory report based from a biology course at California State University San Marcos.|
Keywords: stromatolite, cyanobacteria, origin of life, anza-borrego, bioinformatics, evolution, molecular biology, DNA sequencing
Buchheim, P. (1995). Stromatolites: Living Fossils in Anza-Borrego Desert State Park. https://www.researchgate.net/publication/299610907_Stromatolites_Living_Fossils_in_Anza-Borrego_Desert_State_Park\
Bush Heritage Australia. Stromatolites. https://www.bushheritage.org.au/species/stromatolites
Read, B. Biological Sciences Department. California State University San Marcos. https://www.csusm.edu/biology/index.html
Illumina. Understanding Illumina Quality Scores. Accessed October 14, 2020. https://www.illumina.com/content/dam/illumina-marketing/documents/products/technotes/technote_understanding_quality_scores.pdf
Kothari A., Vaughn M., Garcia-Pichel F. (2013). Comparative genomic analyses of the cyanobacterium, Lyngbya aestuarii BL J, a powerful hydrogen producer. Frontiers in Microbiology Vol. 4 Page 363. https://www.frontiersin.org/articles/10.3389/fmicb.2013.00363/full
Matlock, B. Assessment of Nucleic Acid Purity. Thermo Fisher Scientific. Accessed October 14, 2020. https://assets.thermofisher.com/TFS-Assets/CAD/Product-Bulletins/TN52646-E-0215M-NucleicAcid.pdf
M.D. & Guiry, G.M. (2020). AlgaeBase. World-wide electronic publication, National University of Ireland, Galway. https://www.algaebase.org/search/species/detail/?species_id=23786
Papineau, D., Walker, J. J., Mojzsis, S. J., & Pace, N. R. (2005). Composition and structure of microbial communities from stromatolites of Hamelin Pool in Shark Bay, Western Australia. Applied and environmental microbiology, 71(8), 4822–4832. https://doi.org/10.1128/AEM.71.8.4822-4832.2005
Wikipedia. Melting Curve Analysis. Accessed October 14, 2020. https://en.wikipedia.org/wiki/Melting_curve_analysis
Winner, C. (2013). “What Doomed the Stromatolites?”. Woods Hole Oceanographic Institution. https://www.questia.com/magazine/1P3-3128006171/what-doomed-the-stromatolites