Team:EPFL/Design
Design
Introduction
The aim of the vaccine part is ....
Rising Importance of Cancer Vaccination
Encapsulin
Our Vaccine Design
Library Expression Process
Introduction
The aim of the follow-up part is to provide a proof of concept for detecting disease recurrence, as well as to monitor our treatment efficacy in melanoma patients by detecting specific biomarkers present in the blood. This is particularly important for our project since it constitutes a non-invasive way of validating our vaccine efficacy: tumor biopsies are indeed very invasive, time consuming, and often difficult to perform. Here, we envision a new generation of diagnostic approach, by which a simple liquid biopsy could give us an accurate prognosis regarding the genetic evolution of the tumor in response to our immunotherapy treatment, and would also enable us to detect relapses. This requires a detection system that is both highly sensitive and highly specific, since these biomarkers yield a very precise sequence and are often present in extremely low concentrations in the blood. Our idea to solve this problem is to combine RCA or PCR amplification with a Cas12a-protein based system for a rapid and specific detection. We divided this part in two separate modules, designed to tackle the two different biomarkers we are using: circulating tumor DNA and microRNAs.
Blood Biomarkers
.....miRNAs are instead short (18-24 nt) non-coding RNA molecules which act as post-transcriptional regulators of gene expression. Over the years, miRNAs have been proved to play a critical role in a variety of different diseases and in several aspects of cancer (Larrea et al., 2016). Moreover miRNAs are remarkably stable in human plasma (Mitchell et al., 2008), and recently several miRNAs circulating in the blood have been shown to be dysregulated (either over- or under-expressed) in patients with certain cancers, including melanoma, with respect to healthy subjects (Mirzaei et al., 2016). For these reasons, miRNAs have been proposed as potential prognostic and diagnostic biomarkers for melanoma, which makes them suitable candidates for the follow-up part of our project as well.
Cas12a
To answer the need for a fast and robust detection method we chose to work with the newly characterized Cas12a (Cpf1) protein.
CRISPR-Cas (clustered regularly interspaced short palindromic repeats–CRISPR-associated) system are originally inspired by an antiviral defense mechanism used by prokaryotes which essentially works by recognizing and cleaving the foreign DNA/RNA. It has in the recent years widely been used as a gene editing tool for its ability to find and cut a specific target sequence (the activator).
This activator is composed of two different strands: the target strand (TS) and the non-targeted strand (NTS). The NTS requires a T-rich protospacer adjacent motif (PAM) sequence to be recognized by Cas12a whereas the TS contains the complement sequence of the guide RNA (gRNA), the gRNA being part of the crRNA.
With both these requirements completed, the interchangeable CRISPR RNA (crRNA) will successfully guide the protein to the target.
As a result of cleaving its double stranded DNA (dsDNA) target, Cas12a will undergo a conformational change which will unleash the protein’s endonuclease activity with a single active site in the RuvC catalytic region against any single stranded DNA (ssDNA). This unspecific collateral cleavage is what makes this system so suitable for detection as it greatly amplifies the signal.
In our assays we decided to work with the purified Lba Cas12a (type V-A CRISPR) extracted from Lachnospiraceae bacterium ND2006 and provided by New England BioLabs.
Design of the crRNA
Blablabla
Sample preparation
Sample preparation
Amplification
ctDNA amplification
One disadvantage of a classic CRISPR-Cas based assay is the need to have a PAM sequence near the region that we want to detect, for efficient RNA-guided DNA binding. To eliminate this need, we designed PCR primers that would specifically introduce the PAM sequence, for efficient and sequence-independent detection of any given junction or mutation
Although miRNAs are potentially very valid candidate as biomarkers, they are associated with some hurdles (particularly low abundance) which are not completely overcome by currently existing detection methods (Miao et al., 2015).
Among different recent amplification techniques, Rolling Circle Amplification has been proved to be one of the most suitable, thanks to its robustness, simplicity, specificity and high sensitivity (Cheng et al., 2009). Rolling-Circle Amplification (RCA) is an isothermal amplification (contrarily for instance to Polymerase Chain Reaction) where miRNA (or another short RNA or DNA sequence) is amplified by means of a circular DNA template (i.e. a probe) and a special DNA (or RNA) polymerase: the miRNA acts as a primer, with the RCA product (i.e. the amplicon) consisting in a concatemer containing tens to hundreds of tandem repeats that are complementary to the probe (Ali et al., 2014).
Toehold-initiated Rolling Circle Amplification (tiRCA), in particular, employs phi-29 DNA polymerase and is based on structure-switchable dumbbell-shaped probes (Deng et al., 2014): upon hybridization with the specific target miRNA, one of the two strands of the double-stranded region of the probe is displaced, resulting in an "activated" circular form of the probe with triggers the start of the RCA reaction. The complete mechanism of RCA is shown in the Figure below:
Although it is the probe - and not directly the miRNA - to be amplified, RCA allows to significantly increase the concentration of the miRNA sequence in solution: indeed, since a large portion of the probe is complementary to the miRNA, the amplicon of the probe will incorporate several copies of the original miRNA. This can theoretically be exploited to increase the sensitivity of an assay for quantification of miRNA. As later explained, while our Amplification step was mostly inspired by Qiu et al., 2018, we explored a new, ambitious Detection step after RCA based on Cas12a (and not on Cas9 and split reporter proteins). This implied designing new probes with specific characteristics for Cas12a, as explained in the following sections.
Our probes
The first miRNA we decided to target is let-7a-5p: this miRNA is not among the ones found to be relevant as melanoma biomarkers (as instead are other miRNAs of the let-7 family) (Larrea et al., 2016; Mirzaei et al., 2016); nonetheless, we thought it might be the best option to start from it as a proof of concept, because it was already well characterized for Rolling Circle Amplification (RCA) by Deng et al., 2014 and Qiu et al., 2018
Qiu et al., 2018, as well as our colleagues from the related 2016 iGEM team of NUDT China, had designed their probes in order for the amplicons to be recognized by a CRISPR-Cas 9 system. Since our project deals instead with CRISPR-Cas 12a, despite the miRNA sequence being the same, we therefore had to modify the sequences of our probes accordingly. More specifically, we had to adapt the PAM sequence (placed on the amplicon of the probe) in order to match our Cas protein (we worked with LbCpf1): while the requirement for Cas9 was NGG on the 3' of the amplicon, in our case we needed to have TTTN on the 5'. More details on the design are described in the section "Detailed design".
We wanted to test different designs of probes: some were conceived to have the PAM at the beginning of the larger loop of the amplicon (as in the probes from NUDT China), but we also investigated the case where the PAM was placed on the double-stranded part (the stem) instead; the sequence on the uncostrained large loop was also changed among the probes.
We ordered 10 different probes; the sequence and related notes are described in the Table below.
| Name | Sequence (5'->3') | Description |
|---|---|---|
| Probe 1 | pACAACCTACTACCTCAAACGTAGGTTGTAT AGTTTAAAGGGAGTCGGCGGAACTAT |
Probe designed by our team for Cas 12a. PAM on the large loop of the amplicon. |
Probe 2 |
pACCTCATTGTATAGCCCCCCCCTGAGGTAG TAGGTTGCCCAACTATACAACCTACT |
Probe from Deng et al., 2014 and Qiu et al., 2018 (respectively referred to as "SP-let-7a" and "let-7a probe 1"), designed for Cas9. Used as a control for the efficiency of the amplification. |
| Probe 3 | pACCTCACCCCCCCCCCCCCCCCTGAGGTAG TAGGTTGCCCAACTATACAACCTACT |
Probe from Qiu et al., 2018 ("let-7a probe 2"), designed for Cas9. Used as a control for the efficiency of the amplification. |
| Probe 4 | pACCTCAAAAAAAAAAAAAACCCTGAGGTAG TAGGTTGCCCAACTATACAACCTACT |
Probe from Qiu et al., 2018 ("let-7a probe 3"), designed for Cas9. Used as a control for the efficiency of the amplification. |
| Probe 5 | pACCTCATTTTTTTTTTTTTCCCTGAGGTAG TAGGTTGCCCAACTATACAACCTACT |
Probe from Qiu et al., 2018 ("let-7a probe 4"), designed for Cas9. Used as a control for the efficiency of the amplification. |
| Probe 6 | pACAACCTACTACCTCAAACGTAGGTTGTAT AGTTTAAAGGGGGGGGGGCGAACTAT |
Probe designed by our team for Cas 12a. PAM on the large loop of the amplicon. Large loop with repetitive sequence of Gs. |
| Probe 7 | pACAACCTACTACCTCAAACGTAGGTTGTAT AGTTTAAAGGGAGTGGTTTAAACTAT |
Probe designed by our team for Cas 12a. PAM on the stem. Large loop made of 8 bases. |
| Probe 8 | pACAACCTACTACCTCAAACGTAGGTTGTAT AGTTTAAAGGGAGTCGGCGGTTTAAACTAT |
Probe designed by our team for Cas 12a. PAM on the stem. Large loop made of 12 bases. |
| Probe 9 | pACAACCTACTACCTCAAACGTAGGTTGTATAG TTTAAAGGGGGGGGGGGGGGCGTTTAAACTAT |
Probe designed by our team for Cas 12a. PAM on the stem. Large loop made of 16 bases. |
Probe 10 |
pACAACCTACTACCTCAAACGTAGGTTGTAG AGTTTAAAGGGAGTCGGCGGAACTCT |
Probe designed by our team for Cas 12a. PAM on the large loop of the amplicon. Single base mismatch on the stem with respect to the target miRNA sequence. |
Note: The sequences of the probes include a phosphate group at the 5' end (in order to ligate the probes). We nonetheless always ordered the oligonucleotides without the phosphate (because the cost was significantly lower) and then performed phosphorylation by means of T4 Polynucleotide Kinase prior to ligation.
For each probe we ran an analysis of the secondary structure by means of available servers online (NUPACK, MFold): in all cases the structure of the probe, of its amplicon and of the series of 4-5 copies of the amplicon were tested in order to check the absence of unwanted secondary structures. We also used RNAstructure DuplexFold to test the secondary structure of the dimer probe/miRNA: we were not able to find a more suitable tool for the analysis of the duplex; nonetheless we believe that this server, despite its limitations with respect to our analysis (no possibility of having a circular probe, no possibility to have a DNA/RNA dimer), was enough to show qualitatively the interaction between our probe and let-7a.
SYBR Green I
Two main alternatives are suitable in order to test the efficacy of Rolling Circle Amplification (Deng et al., 2014; Qiu et al., 2018). First of all, the amplicons can be tested by means of an agarose gel to verify the size; nonetheless, this method shows some limitations because of the large size of the amplicons. Indeed, as we also saw from our experiments (link to the Notebook), the size of the amplicons after a 2 hour-RCA is so large that the band is extremely close to the gel.
A more valid alternative is instead to perform a real-time fluorescence measurement by means of SYBR Green I.
SYBR green I is an intercalating dye that preferentially binds to minor grooves of double-stranded (dsDNA) (Zipper et al., 2004). It has also been shown to bind to single-stranded DNA (ssDNA) and RNA (for which instead SYBR Green II is a more suitable option (Sigma-Aldrich)), but with a significantly lower performance (Vitzthum et al., 1999).
When complexed with nucleid acid, SYBR Green I absorbs blue light (maximum excitation wavelength is 497 nm) and emits green light (emission peak at 520 nm) (Sigma-Aldrich), which makes it suitable for quantification - by means of a plate reader - of the DNA amplicons (i.e. the reverse complement of the probes) from our Rolling Circle Amplification (RCA).
Indeed, since we verified in all cases the absence of unwanted secondary structures (more details in Detailed Design), the stems in the probes and in the amplicons are the only double-stranded targets to which SYBR Green I can preferentially bind: this allows to observe the increase over time in the size of the amplicon during RCA.
Detailed design
This section is more specifically dedicated to the reasonings behind the sequences of our probes.
Analysis of given probes
We started our design from the analysis of one probe from Qiu et al., 2018, namely "let-7a probe 1" (Probe 2 for us). The sequence was the following one:
5’-pACCTCATTGTATAGCCCCCCCCTGAGGTAGTAGGTTGCCCAACTATACAACCTACT-3’
where:
- the regions in italic are those belonging to the loops of the hairpin
- the regions in orange and green are those belonging to the stem of the hairpin (and which are complementary with each other)
- the underlined region is the one complementary to the miRNA (let-7a-5p: UGAGGUAGUAGGUUGUAUAGUU)
Such probe consists of a double-stranded stem part, a 10 bases-long loop (which from now on we will refer to as "small loop" - on the right in the figure above) and a 16 bases-long loop ("large loop" - on the left). As we can observe, the toehold region of the probe (i.e. the part on the small loop where the miRNA binds) is 7 bases long, in accordance with Deng et al., 2014, who proved it to be the optimal length to achieve both sensitivity and specificity.
The amplicon of such probe is therefore*:
5'-AGTAGGTTGTATAGTTGGGCAACCTACTACCTCAGGGGGGGGCTATACAATGAGGT-3’
where:
the sequence in bold is the one which is complementary to the gRNA (except for two mismatches, which are highlighted) and the region in red is the PAM sequence (in this case single stranded).
We emphasize here that the PAM sequence is on a single-stranded part of the amplicon (the one complementary to the large loop of the probe): therefore, such single-stranded PAM can only be present on the amplicon, and not on the probe itself (as would have been instead if the PAM was on a double stranded part).
The gRNA sequence (as indicated by Qiu et al., 2018) is:
5’-ACUGUACAAACUACU|ACCUCA(GUUUUAGAGCUAGAAAUAGCAAGUUAAAAUAAGGCUAGUCCG) -3’
with the scaffold region indicated in parentheses. The region out of the brackets is the spacer, binding to the amplicon, and the sequence in italic corresponds in particular to the part of the spacer binding on the loop of the amplicon (with the rest of the spacer binding to the stem). The sign | indicates the position where the gRNA binds to the point on the amplicon where each new "copy" of the amplicon is considered to start (i.e. the point where the 3' of a "subunit" of the amplicon and the 5' of the successive subunit are linked together).
More specifically, we can notice that in this design the spacer coincides with the reverse complement of let-7a, with the exception of the two mismatches and of a missing A at the beginning. The template of the gRNA for Cas9 would therefore be:
5'-[reverse complement of miRNA]-[scaffold]-3'
The expected interaction between amplicon and gRNA is outlined in the figure below:
We can observe how the PAM sequence (in red in the figure) is located at the very beginning of the large loop in the amplicon, whereas the gRNA binds to the whole stem part and partially to the small loop.
*Here and after, when referring to the "amplicon sequence", we only show one single copy of the reverse transcript of the probe. The actual amplicon, by definition of Rolling Circle Amplification, is of course made instead of sequential copies of this "unitary" sequence.
Comparison Cas9/Cas12a
We then tried to design our own probes for Cas 12a, working backwards from the gRNA.
Contrarily to Cas 9, for which the PAM must be on the 3' side of the target, for Cas12a the PAM must be on the 5’ side of the target instead. This implies that the scaffold part of the gRNA must be on the 5’ side (instead of the 3’) as well (Figure below).
Below is shown a direct comparison of the interaction between target amplicon and gRNA for Cas 9 and Cas 12a.
We therefore conclude that the template for our guide RNA for Cas 12a should be:
5’-(UAAUUUCUACUAAGUGUAGAU)NNNNNN|NNNNNNNNNNNNNNNN-3’ [gRNA template]
where the sequence in parentheses indicates the scaffold of the gRNA for LbCas12a. The sequence out of the brackets is the spacer, binding to the amplicon, and in particular the sequence in italic corresponds to the part binding on the loop of the amplicon.
The spacer is therefore 22 bases long (as let-7a-5p), 15 of which bind to the stem part of the amplicon and the remaining 7 bind to the small loop of the amplicon. Note that the gRNA for Cas9 from Qiu et al., 2018 was instead 21 bases long (15 and 6): we decided to add one more base at the end to completely match the length of the miRNA.
We can notice that also in this design the spacer has to coincide with the reverse complement of let-7a (as for Cas 9) . The template of the gRNA for Cas12a would therefore be:
5'-[scaffold]-[reverse complement of miRNA]-3'
Probes for Cas12a
From the template above we can therefore conclude that the gRNA for our Cas 12a system, designed as the one for Cas 9 from Deng et al., 2014, has to be:
5’-(UAAUUUCUACUAAGUGUAGAU)AACUAU|ACAACCUACUACCUCA-3’ [gRNA sequence]
From the specifications for the probe above (10 bases small loop, 16 bases large loop) and from the gRNA sequence, the template amplicon therefore needs to have the following structure:
5’-ATAGTTNNNNNNNNNNNNTTTNAACTATACAACCTACNNNTGAGGTAGTAGGTTGT-3’ [amplicon template]
We then proceeded to define the bases for the Ns, aiming not to have unwanted minor secondary structures (e.g. smaller loops) in the loops. This was done mostly by considering pairing principles, e.g. avoiding non-Watson-Crick interaction (e.g. T-G) which might be thermodynamically favoured or trying not to have complementary bases with more than 1 base in between (which might lead to hairpin loops). In all cases, the minimum free energy structure (MFE) was plotted by means of the available software (NUPACK, Mfold), both for the amplicon and the probe - i.e. its reverse complement-, to check that the intended dumbbell shape was indeed achieved.
This lead us to the sequence of Probe 1 and Probe 6 (Probes from 2 to 5 were the probes for Cas 9 from Deng et al., 2014 and Qiu et al., 2018).
-----
We also wanted to test the case of probes having the PAM sequence not on the large loop, but on the stem instead (i.e. a double-stranded PAM, as usually required in Cas systems, and not single-stranded). We considered in this case three different alternatives:
Changing 4 bases in the large loop in order for them to be complementary to the PAM sequence, without adding more bases. This leads to a 19 bases-long stem, a 10 bases-long "small" loop and a 8 bases-long "large" loop. The template sequence of the amplicon is the following one:
5’-ATAGTTN'AAANNNNNNNNTTTNAACTATACAACCTACNNNTGAGGTAGTAGGTTGT-3’ (with N' being the base complementary to the N in the PAM)
Inserting 4 more bases complementary to the PAM on one end of the large loop (after ATAGTT), without changing any base. This leads to a 19 bases-long stem, a 10 bases-long small loop and a 12 bases-long large loop. The template sequence of the amplicon is the following one:
5’-ATAGTTN'AAANNNNNNNNNNNNTTTNAACTATACAACCTACNNNTGAGGTAGTAGGTTGT-3’
Inserting 4 more bases complementary to the PAM on one end of the large loop (after ATAGTT) and 4 more bases at the other end of the large loop (before the PAM sequence), in order to keep the original length of the large loop (16 bases). This leads to a 19 bases-long stem, a 10 bases-long small loop and a 16 bases-long large loop. The template sequence of the amplicon is the following one:
5’-ATAGTTN'AAANNNNNNNNNNNNNNNNTTTNAACTATACAACCTACNNNTGAGGTAGTAGGTTGT-3’
These three alternatives led, respectively, to Probe 7, Probe 8 and Probe 9.
-----
Finally, Probe 10 was designed in a way to have a mismatched base in the stem with respect to the let-7a sequence (highlighted in both strands below):
5'-pACAACCTACTACCTCAAACGTAGGTTGTAGAGTTTAAAGGGAGTCGGCGGAACTCT-3'
Design of gRNAs after new theory
Halfway through our project (see Notebook for more details), after starting testing our amplicons with Cas12a and the fluorescent reporter (DNase Alert), we realized that the probe itself (more specifically the product of RCA in the absence of miRNA, i.e. with no amplicon) was triggering the Cas system causing a very high fluorescence signal, comparable to the signal obtained for the samples with miRNA (i.e. with probe+amplicon).
We hypothesized that this was due to the fact the our Cas12a was working PAM-independently (more details in "New theory on Cas12a activation - miRNA" in the "Fluorescence readout" section). More specifically, our gRNA was meant to target the whole stem (and in addition 7 bases in the small loop) of the amplicon; since the stem is double-stranded, the target sequence for the gRNA is also present in the probe (in the opposite strand).
This would not have been a problem if the Cas had been working, as expected, PAM-dependently, because the PAM is only contained in the amplicon, not in the probe. Nonetheless, if the Cas does not need the PAM sequence, but simple recognizes a target from the sequence of the gRNA, then also the probe itself is recognized as a target. Moreover, since the concentration of the probe in the RCA reaction is higher than the expected concentration of amplicon, the signal from the probe behaves as noise, overcoming the signal of interest (i.e. from the amplicon).
We therefore designed a new guide RNA with the aim of targeting only the amplicon and not the probe. Our idea was to have the gRNA binding not on the stem, but on the large loop of the amplicon instead. Since the loops of the amplicon are single-stranded (and not double-stranded as the stem) this should allow the gRNA to target only the amplicon and not the probe, being the target sequence contained only in the amplicon and not in its reverse-complement: more specifically, we decided to design a guide RNA perfectly complementary to the large loop of the amplicon of Probe 1; in this way Probe 1, having on the contrary exactly the same sequence as the gRNA, should have never been targeted by this new gRNA.
As from the template gRNA above (5'-[scaffold]-[reverse complement of miRNA]-3'), the spacer was therefore modified to bind (with perfect match) to the large loop of the amplicon of probe 1.
Two different designs were tested, one - referred to elsewhere as "S_1" - binding to the whole large loop and to the first 4 bases after the large loop (for a total of a 20 bases-long spacer), and one - "L_1" elsewhere - binding only to the large loop (16 bases-long spacer). The complete sequences are the following ones:
5'-(UAAUUUCUACUAAGUGUAGAU)UAAAGGGAGUCGGCGGAACU-3' [gRNA sequence - S_1]
5'-(UAAUUUCUACUAAGUGUAGAU)UAAAGGGAGUCGGCGG-3' [gRNA sequence - L_1]
The comparison between the mode of action of the previous, original gRNA and the "new" ones is better explained in the figure below:
Fluorescence readout
ctDNA Cas assay
One disadvantage of a classic CRISPR-Cas based assay is the need to have a PAM sequence near the region that we want to detect, for efficient RNA-guided DNA binding. To eliminate this need, we designed PCR primers that would specifically introduce the PAM sequence, for efficient and sequence-independent detection of any given junction or mutation
miRNA Cas assay
References
- Mitchell, Patrick S., et al. "Circulating microRNAs as stable blood-based markers for cancer detection." Proceedings of the National Academy of Sciences, 105.30 (2008): 10513-10518.
- "EnGen Lba Cas12a (Cpf1)" - New England BioLabs website. URL: https://international.neb.com/products/m0653-engen-lba-cas12a-cpf1#Product%20Information_Notes (Accessed 24/09/2018)
- Miao, Peng, et al. "Ultrasensitive detection of microRNA through rolling circle amplification on a DNA tetrahedron decorated electrode." Bioconjugate chemistry, 26.3 (2015): 602-607.
- Cheng, Yongqiang, et al. "Highly sensitive determination of microRNA using target-primed and branched rolling-circle amplification." Angewandte Chemie International Edition, 48.18 (2009): 3268-3272.
- Ali, M. Monsur, et al. "Rolling circle amplification: a versatile tool for chemical biology, materials science and medicine." Chemical Society Reviews, 43.10 (2014): 3324-3341.
- Zipper, Hubert, et al. "Investigations on DNA intercalation and surface binding by SYBR Green I, its structure determination and methodological implications." Nucleic acids research, 32.12 (2004): e103-e103
- "SYBR Green II RNA Gel Stain" - Sigma-Aldrich. Datasheet. URL: https://www.sigmaaldrich.com/content/dam/sigma-aldrich/docs/Sigma/Datasheet/2/s9305dat.pdf (Accessed 11/10/2018)
- Vitzthum, Frank, et al. "A quantitative fluorescence-based microplate assay for the determination of double-stranded DNA using SYBR Green I and a standard ultraviolet transilluminator gel imaging system." Analytical biochemistry, 276.1 (1999): 59-64.
- Larrea, Erika, et al. "New concepts in cancer biomarkers: circulating miRNAs in liquid biopsies." International journal of molecular sciences, 17.5 (2016): 627.
- Mirzaei, Hamed, et al. "MicroRNAs as potential diagnostic and prognostic biomarkers in melanoma." European journal of cancer, 53 (2016): 25-32.
- "SYBR Green I nucleic acid gel stain" - Sigma-Aldrich. Datasheet. URL: https://www.sigmaaldrich.com/content/dam/sigma-aldrich/docs/Sigma-Aldrich/Datasheet/s9430dat.pdf (Accessed 11/10/2018)
- Deng, Ruijie, et al. "Toehold-initiated rolling circle amplification for visualizing individual microRNAs in situ in single cells." Angewandte Chemie, 126.9 (2014): 2421-2425.
- Qiu, Xin-Yuan, et al. "Highly Effective and Low-Cost MicroRNA Detection with CRISPR-Cas9." ACS synthetic biology, 7.3 (2018): 807-813.
- Zadeh, Joseph N., et al. "NUPACK: analysis and design of nucleic acid systems." Journal of computational chemistry, 32.1 (2011): 170-173.
- Zuker, Michael. "Mfold web server for nucleic acid folding and hybridization prediction." Nucleic acids research, 31.13 (2003): 3406-3415.
- Reuter, Jessica S., and David H. Mathews. "RNAstructure: software for RNA secondary structure prediction and analysis." BMC bioinformatics, 11.1 (2010): 129.
- Xie, Kabin, and Yinong Yang. "RNA-guided genome editing in plants using a CRISPR–Cas system." Molecular plant, 6.6 (2013): 1975-1983.