Boy, it's been a while since my last blog post. Unfortunately, my 6-year-old computer has been really unstable on certain sites and crashes whenever I try to edit the site, so the blog has been on hiatus for about a year now. BUT I've managed to get my hands on my sister's old Chrome Book, so hopefully, I'll be breathing new life into the blog.
It would take too long to update you on all my activities since my last post, but I'll mention that a highlight of the last year has been the premiere of my new string quartet: Syadvada at the 2016 SoundSCAPE Festival in Maccagno, Italy! I'm also working with some excellent Ithacans to bring the piece to life on this side of the Atlantic. I met some incredible musicians at the festival, and I'm incredibly grateful to have been able to add to the conversations that were had there.
Rather than reminiscing about my other activities this past year, I thought I'd talk about a piece I'm writing now: Cicada for String Quartet and Oboe, specifically the fourth movement, whose musical material is derived from the DNA sequence of a bacterium! In this post, I want to talk about the various processes I'm using to decode the various combinations of G, A, T, and C into music. This article will assume a basic foreknowlege of genetic biology, as providing that would go beyond the scope of a simple blog (yet still admittedly long) post. BUT, if you need to refresh your High School Bio chops, this vid, this vid, and others in the same series should give you a crash course (I certainly needed to re-learn a few things before embarking on this project).
I'll briefly give you an overview of the work, then we'll focus on just the fourth movement. Jake is an avid fan of entomology—indeed, many of his own works are named after bugs. I knew when he apporached me about writing him a piece, it was gonna be about bugs. I decided to write the piece about Cicadas—they're loud and melodious animals, with rich poetic associations in many different cultures and literary traditions. I could begin to see connections between this insect and many other disciplines, and, seeking to unite them through music, I began to write the piece. The work is composed of six movements that represent a journey of an organism evolving from primordial crystalline materials into rich organic complexity:
I. Cicada (a prologue)
II. Ocean (cyclical, primordial)
III. Impluse (an organism coming to life, awakening)
IV. Symbiosis (growing rapidly, self-generating)
V. Heartbeat (a complex organism, animal)
VI. Sentience (conscious, self aware)
Note that the parentheses aren't part of the movement titles.
Something I like about being a composer is that it gives you an excuse to deeply investigate anything. — Kate Soper
Nutrient Synthesis from DNA
When an organism needs to synthesize a complex molecule like Cobalamin or Hemoglobin, it often takes an "assembly-line" of proteins working together one after the other to take simpler molecules and put them together to form bigger ones. In other words, if you wanted to make hemoglobin, for example, you wouldn't be able to code a single gene for it. You'd need several genes that produce a series of proteins that all work together to create it (especially in the case of hemoglobin, because it's really a combination of several smaller proteins folded together). Another great example is Serotonin synthesis from Tryptophan which you can see here has a simple multi-step metabolic pathway.
Cobalamin—aka Vitamin B12—has a 13-step metabolic pathway as it is synthesized in C. H. cicadicola. Cobalamin is a vital nutrient for all animals and is only synthesized in bacteria, not in plants or other animals. Though C. H. cicadicola has one of the shortest known genomes, only around 150,000 base-pairs long, there was no WAY I was gonna be able to turn its entire genome into music—it was just too much information. Instead, I decided to select a gene or set of genes that represented the symbiotic relationship the bacterium has with the cicada. The genes involved in Cobalamin synthesis were a pretty good option, though there were plenty more. It wasn't until I learned about the second level of symbiosis that was going on in some species of cicadas of the genus Tettigades that I made my final decision.
Sympatric Speciation and Second Order Symbiosis
As I was researching the actual biological mechanisms that were involved in this symbiosis, I found an incredible recently-published paper, Sympatric Speciation in a Bacterial Endosymbiont Results in Two Genomes with the Functionality of One, by James T. Van Leuven, Russell C. Meister, Chris Simon, and John P. McCutcheon. The paper discusses how in Tettigades undata, a species of cicada, the C. H. cicadicola has actually speciated into two bacteria with complemetary genomes! To understand how awesome this is, let's briefly discuss sympatric speciation. To quote Wikipedia, "Sympatric speciation is the process through which new species evolve from a single ancestral species while inhabiting the same geographic region." In other words, this ONE species of bacterium has split into TWO distinct species of bacteria inside the cicada! We can refer to these two forms of Hodgkinia by the names they are given in the paper, TETUND1 and TETUND2 (for Tettigades undata 1 and 2).
A diagram below, taken from the paper, shows what I just explained for the genes that code for the proteins in the assembly-line that produce cobalamin (B12) in the different versions of C. H. cicadicola. The diagram on the left with the green circles shows the genes present in the DNA of TETULN (the "normal" Hodgkinia) that code for proteins in the assembly-line. On the right, we see the same assembly line, but some of the genes in the sequence are only in the TETUND1 genone (orange) and some are only in the TETUND2 genome (blue). If you cut out the genes in either of the sequences, it would break the chain in several places and all the organisms involved would perish. So we can see how not only are the Hodgkinia and host insect codependent, but, in T. undata, each version of Hodgkinia, TETUND1 and TETUND2, are also codependent, giving us a multi-level symbiotic relationship.
Is your mind blown? Good. Now let's make some music.
Bioserialism: Decoding the Genome into Music
With my gene sequence chosen, the only problem left was turning this into music! Not an easy task. Each gene in the sequence is made up of hundereds if not thousands of base-pairs. I needed a way of translating Gs, As, Ts, and Cs into useable information. If only there were a way... hmmm... OH WAIT your body and every other living thing on the planet does this every DAY through the 4 billion year-old process of genetic translation!
The videos linked above do a great job at explaining genetic translation. Let's work it backwards. In a nutshell, proteins are just long chains of smaller molecules called amino-acids. There are twenty-or-so kinds of these amino acids that are proteinogenic (i.e. that are put together into proteins from the information in your DNA). Depending on which of these amino acids link up, how many, and in what order, the resultant protein's structure is defined. There's an organelle inside the cell called a ribosome. This is a little machine responsible for linking together the different amino acids depending on the information that gets fed into it in the form of messenger RNA. This mRNA is essentially a photographic negative of a gene's nucleotide sequence from the DNA. As it's fed into the ribosome, the mRNA is read three nucleotides at a time. Each of these nucleotide triplets (of which there are 4^3 combinations, so 64) is called a codon. The ribosome will add a different amino acid to the chain depending on which codon it's reading. As the mRNA passes through the ribosome, the ribosome reads it codon-by-codon and links together the appropriate amino acids to form proteins.
That's obviously a watered down version of how translation works, but that's the general idea. So now we know that each of the 64 possible codons must somehow map onto each of the twenty-or-so proteinogenic amino acids. There are a bunch of ways you could do this, but after 4 billion years, Life on Earth has settled, for the most part, for this:
Our problem of mapping nucleotides onto musical material has been solved! We don't need to map specific nucleotides onto music. We don't even need to make our own musical codons (although you could...). We just need to map musical parameters onto the 20 or so proteinogenic amino acids and let nature's 4 billion year-old translator take it from there! Out of the raw genetic code, we can generate musical proteins that twist, fold, and interact with one another exactly like their biological counterparts.
Let's not get too far ahead of ourselves, however. The next question becomes this: how do we decide WHICH musical parameters to map onto which amino acids? We could simply decide arbitrarily that certain amino acids map to certain musical notes or other parameters, but I prefer a more finessed approach, one that's informed by the physical and chemical properties of each individual amino acid.
We're about to go full orgo in here. Prepare yourself.
Bioserialism: Music from Molecular Structures
If both the carboxyl and amine groups are bonded to the first (or alpha-) carbon, the amino acid is called an alpha-amino acid. All of the proteinogenic amino acids are alpha-amino acids, and so they have exactly the same molecular structure, differing only in the composition of their side chain. SO, the properties of the amino acid depend on what its side chain looks like.
To the right is a great diagram of an alpha amino acid via Wikipedia. Note the central alpha-carbon that acts as the hub to which everything else connects. The NH2 group that connects to it on the left is an amine group. The COOH group that connects to it on the right is a carboxyl group. A lone Hydrogen sits perched on top. The boxed R on the bottom represents where any side chain would connect, defining the properties of the amino acid. I should mention that this model is geometrically oversimplified. The picture makes the molecule look like it lays flat. It doesn't. The atoms and groups of atoms in the molecule stick out in three dimensions, and can actually spin around their single bonds.
So if we want to understand the properties of each proteinogenic amino acid, we need to familiarize ourselves with the compositions of their side chains. Below is a table—also from Wikipedia—that diagrams all 20 porteinogenic amino acids (plus selenocysteine, but ignore selenocysteine) and their side chains. The table sorts the acids into groups depending on their polarities and such, but for now just pay attention to the patterns in the physical structures in the different side chains.
Drawing the chemical structures of each of the side chains out on paper helped me a lot to internalize their structures and begin to see patterns. Check out the bottom right-hand side of the first page of my sketches above. Glutamine, Glutamic Acid, Asparagine, and Aspartic Acid all have extremely similar structures. Glutamine and Asparagine are basically the same except for the extra methylene bridge that Glutamine has, both ending in a carboxamide group. Same deal with Glutamic Acid and Aspartic Acid. Glutamic Acid has exactly one more methylene bridge than Aspartic Acid, and instead of ending in carboxamide groups, they end in carboxyl (COOH) groups. Reading up on their chemical properties, we find that all four of these moluecules behave somewhat similarly. We can therefore give these amino acids similar musical funtions—such as mapping them to adjacent pitch classes. Digging deeper into the chemical functions of these molecules, we learn that Aspartic Acid and Glutamic Acid in particular are often paired with Arginine and Lysine in proteins. Argenine and Lysine have similar protein structures: long chains of methylene bridges with some kind of nitrogen-bearing funcional group on the end (Guanidine for Arginine, a simple Amine for Lysine). Because of these similar stuctures, they bear similar chemical functions. Armed with all this information, we can then associate Arginine and Lysine with adjacent pitch classes that bear some relation to the pitches represented by Aspartic and Glutamic Acid.
Let's take a look at our remaining amino acids; there are some more relationships we can draw! For example, Serine and Threonine both end in hydroxyl groups, giving them similar chemical functions—they're both quite hyrophillic, aiding in water soluability. On the other hand, the side chains of Leucine, Isoleucine, and Valine are all made of methylene groups bonded together in different ways. Leucine and Isoleucine can be grouped together, as have the same number of atoms but with different geometric structures. Valine has one less methylene than Leu or Ile, but still functions quite similarly. These molecules function hyrdophobically, so we can make a big group out of all three of them. Living in a more neutral realm between hydrophobia and hydrophilia is Alanine, which is made of a simple methyl group. We can associate this amino acid with a pitch class that lives somewhat neutrally between those given by the previous hydrophobic and hydrphillic groups.
We won't focus just yet on defining explicit pitches for the aforementioned amino acids; what's important is the relationships between them that are analogous to the chemical relationships betweem the molecules. We can express these relationships in different ways. For example: we can map groups of amino acids to pitches that lie in the same overtone series, we can cluster them together as mentioned before, we can set them apart by tritone (alluding to the tritone-sub relationship found in Jazz), etc. In Symbiosis, I used all of these methods to illustrate relationships between pitch classes and amino acids.
The astute reader will notice that we still have eight amino acids that have yet been unassigned a musical function. These amino acids are weird. They have weird atoms and functional groups in them, are weirdly shaped, or behave quite uniquely. These amino acids are responsible for modifying the notes that the other amino acids produce. I've divided these modifiers into two groups: static modifiers and dynamic modifiers. Static modifiers are switches. When we read one of these amino acids, it switches some musical parameter on or off, like a filter. This behavior is linear as we read down the amino acids in a protein. Dynamic modifiers are more complex, and affect the music in a non-linear way. These modifiers can go back and change entire sections of music that were already produced linearly by the machine. More on these later. Let's talk about the molecules.
Methionine is a funny little amino acid. For a few reasons. First, it is always the first amino acid in any protein chain (although it can sometimes get abscised later). This means that all of our 13 cobalamin proteins are going to start with Methionine. Bear that in mind. Second, it has a Sulfur atom in it! Like what!? Sulfur likes to bond with itself. A lot. As it happens, Cysteine also contains sulfur in its side chain, and often bonds with itself in various proteins. Cysteine's disulfide bonds help to strengthen the structural integrity of proteins (common in structural proteins such as Keratin). Thus we can make Cystine stregnthen our musical dynamic, and Methionine, being located mostly at the beginning of amino acid chains, can act as a reset. In other words, Cysteine crescendos, Methionine diminuendos.
The side chain for Glycine is the most simple of the 20 proteinogenic amino acids, composed only of a single hydrogen atom. Glycine is common in collagen, a main component of connective tissue, and can add structural flexibility to proteins. We can use it to connect notes together in our music, or switch between legato and staccato and arco and pizzicato.
Histidine has a funky imidazole ring on its butt. The nitrogen in this ring allows histidine to behave differently in different environments depending on acidity. Its behavior regulates the conformation of the protein it's a part of, allowing the protein to change how it folds and bends. We're going to give histidine a regulatory role in our music as well, allowing it to toggle changes in tempo and rhythm. In Symbiosis, Histidine toggles the "metric atom" between an eighth note and a dotted eighth. The eigth-note atom is the fast tempo, the dotted eighth is the slow tempo (see musical example below).
Look at Proline for a sec. It's montrous. It's like... a Frankenamino! It's cannibalizing itself, its pyrrole side chain bonding back onto the original alpha-carbon. Scary stuff. Because of its quazimodoesque nature, Prolines often create abrupt kinks in peptide chains. For our purposes, Proline will trigger color-shifts making us change timbres or instruments as we grow our musical proteins.
The remaining proteins contain aromatic phenyl or indole groups—vital components of countless organic compounds—and are all naturally fluorescent. For this reason, we're giving them special treatment. These will be our dynamic modifiers, special modules that can act non-linearly during our compositional process. Tryptophan's indole group is made of a benzene ring that's curled back in on itself, fusing with a pyrrole ring. In our piece, Tryptophan will retrograde pitches up until the previous dynamic modifier. Tyrosine and Phenylalanine can be converted into each other through the addition or subtraction of a hydroxyl group. Thus they will both be given the ability to invert the music that is generated. For Phenylalanine, however, we will also impart it with the ability to retrogade the music. In summary, Trp gives you the retrograde, Tyr gives you the inversion, and Phe gives you the retroversion of the previous amino acids in the chain up until the last dynamic modifier.
Now the process is simple: run the amino acids in our proteins through our translation machine and transcribe the music that results! Simple, right? It certainly sounds simple. But no. It is not simple. It is arduous and tedious and challenging in a thousand different ways.
Bioserialism: Transcription and Composition
Transcribing the amino acids into music is relatively easy, albeit tedious. We essentially go through each of the amino acids in each of the proteins in the cobalamin sequence, staff paper in hand, and write the appropriate music, note by note, dynamic by dynamic, etc. To the right, you can see a list of the first six porteins in the cobalamin series, plus their constituent amino acids, abbreviated using standard conventions.
When I first apporach transcribing a protein, I first go through it and do a few things. First, I highlight every Histidine in the sequence. These represent tempo changes, and they act as structural pillars in the resultant musical protein.
Then I go through and circle every Proline. Remember, Proline's molecular structure is kind of cannibalizing itself, creating kinks in real proteins. In the music, it triggers instrument changes or sharp changes of color. These also act as structural pillars in the piece, and can be very helpful in orienting oneself when coming back to work on the piece after a few days.
In the list to the right, you'll also notice that there are some amino acids that have been bolded. These represent amino acids that are coded by parts of genes that overlap each other on the DNA strand. Remember, DNA is made of two chains of nucleotides, so you can have one gene on one of the chains going one direction, and one chain on the other one of the chains going the other direction, and they can overlap. If that's hard to picture, don't worry about the specifics for now. Just remember that the bold represents overlap. When we begin writing musical proteins in paralel with one another (yikes), the overlapping sections of adjacent genes will determine the musical overlap of the contrapuntal proteins.
After all this prep work, I go letter by letter transcribing the music, underlining the amino acids as I go to keep my spot. In the image above, you can see that I originally tried to underline the amino acids on the computer. That proved to be too tedious, however, so I printed the page out and did it by hand.
If we just try to run the proteins through our musical translator and keep whatever comes out, we run into several issues. First, the music must be played by real musicians, so if we end up with something unplayable—like consecutive pizzicati at 100 miles per hour or a giant legato oboe slur with no space for breaths... we're gonna have a bad time. Second, what comes out of "the machine" will just be a stream of notes at various dynamics, tempi, and articulations, played by various instruments. If we want to convey a sense of musicality and drama throughout the piece, we need to turn that stream of notes into real art. Both of these issues can be solved in a variety of ways that involve all taking the rules that we've painstakingly set up for ourselves and bending or breaking them in ways that make artistic sense.
To avoid unplayable passages, we need to invoke the very concept that drives this piece at its core: Symbiosis. If we take the stream of notes in a single protein and explode them across the ensemble, hocketting the notes as needed, we can not only solve our unplayability issues (giving string players time to pizz and our oboist time to breathe), it also creates a richer musical texture and embeds metaphor into the musical relationship between the players. They all need each other to "synthesize" each of the musical proteins. If one fails, they all fail. Symbiosis. This symbiosis between players is inherent in any ensemble setting, but I think the hocketting nature of the work highlights that relationship even more.
So I did. :)
Bioserialism: Growing a Multilevel Musical Protein Structure
The secondary structure has to do with how small segments of the protein twist and curl up around themselves. There are generally two main shapes into which small segments arrange themselves: alpha-helices and beta sheets. In alpha-helices, the amino acids twist around each other to create long spirals. In addition to the image to the left, the image of hemoglobin above shows many of these spirals. In beta-sheets, adjacent strands of a protein layer themselves next to one another through hydrogen bonding.
In our music, we can reflect this secondary structure in several ways. By hocketting the notes across the ensemble as described above, we get a kind of spiraling motif, evoking the structure of the alpha-helix. Then, by doubling instruments with the main musical line in certain spots, we can "hydrogen bond" different layers of the same musical protein to each other to create musical beta-sheets.
So how do we decide where in the music to create these secondary structures? Remember, this structure is hierarchical, so we decide where to form these secondary structures in the music based on the musical primary structure. That is to say, the music itself dictates its own structure organically as it grows (its neither prescribed by the composer nor based on the real-world secondary structure of the proteins in the cobalamin sequence).
When we need to hocket a section between instruments, that creates a musical alpha-helix. For beta sheets, it's slightly more involved. As we transcribe the amino acid sequence into music, we can find patterns emerge: notes that repeat themselves, chromatic background lines, groups of notes or contours that consistently show up, etc. As artists, we can accentuate these macrosopic patterns through the growth of musical beta sheets, i.e. while one instrument plays the primary structure (the stream of notes), the other instruments in the ensemble double it in particular spots, throwing the macroscopic patterns into stark relief. This allows us to grow actual phrases of music from the stream of notes in which repeating patterns spin out of the texture, structural notes are targeted over a long range, and musical events are triggered by other musical events! Suddenly the music is talking to itself.
The musical examples below illustrate musical alpha spirals and beta sheets in greater detail.
Tertiary structure is the 3D shape of an entire contiguous protein chain. This is defined by how local areas of the protein (i.e. alpha-helices and beta-sheets) fold and bond with each other. In the music, we can show this by layering the hocketting alpha spirals with the macroscopic beta-sheets (see examples below). Bear in mind that this is all happening in the same contiguous musical protein. If we want to show quaternary stucture, we're going to need to go further.
This is where it gets. Insane.
A protein's quartenary structure is defined by how smaller contiguous sub-proteins join together to form a large protein complex. Remember how we saw that hemoglobin was actually a combination of smaller sub-proteins folded together? Same thing going on. In our music, we can show this quaternary structure by layering distinct proteins in our cobalamin sequence on top of one other contrapuntally. You heard me. This is where things get NUTS, especially because different musical proteins have different Histidines in different places, meaning they'll switch tempi independently of one another. Yummy. As mentioned before, we overlap the proteins based on where their genes overlap in the Hodgkinia DNA. Composing parts of Symbiosis with overlapping proteins is the most difficult part of the process, since we need to be aware of the counterpoint and the polytempo, and there are fewer instruments to use if we need to create alpha spirals. Still, it's extremely satisfying to find instances where two or more musical proteins line up or play briefly in unison; these delicious moments of consonance represent the hydrogen bonding by which different sub-proteins bind to form larger protein complexes. See the examples below for a taste of how this works.
Concluding Thoughts: What I've Learned as a Composer
The process of bioserialsim (yes I just made that word up, but it sounds cool so I'm keeping it) and how it relates to other compositional processes is something I've thought a lot about. The way we mapped amino acids onto musical parameters is similar to creating a Tone Row, but it differs in that we don't really have control over the ordering of the row; that is left up to nature. It's funny, while the ideas behind this piece may on the surface seem quite academic, in the end it's quite firecely Romantic. After all, the music is on the deepest level about a relationship with nature, one of the cornerstones of Romanticism. The music diverges from old Romantic thought, however, in that it uses modern technology to examine natural processes on a granular level instead of just marvelling in awe over their macroscopic beauty. I think this speaks a lot to the spirit of the times, since the technique of Spectral Analysis—a very popular technique nowadays—is essentially doing the same thing: taking a natural process (e.g. the sound of a gong) and breaking it down analytically to make music out of its parts.
Thinking about this technique of art from the granular analysis of natural processes makes we wonder how else it could be applied. As we mentioned, acoustic spectral analysis is one form of this. One could also perform spectral analyses on non-acoustic spectra such as atomic emmission spectra, brain waves, or ocean waves, or even the patterns in the rings of Saturn, for example. Xenakis made music out of stochastic processes that were mathematical models for real-world processes, but with current technology it is possible to create music from realworld stochastic systems in real time. For example, you could put accelerometers on the nodes of a double pendulum and as that pendulum swings chaotically, information could be sent from the accelerometers to a Max or SuperCollider program that made music in real time. Another idea is to link different multidimentional systems together, mapping their dimensions onto each other using matrices. So for example, you could have a motion traker on a performer that controls electrical currents running through different incandescent materials, and as those material incandesce, you analyze their emission spectra in real time and map that onto music that's produced electronically.
Besides the analytical approach I took on this piece, I've also had the chance to explore other musical techniques. The granular, multi-level structure of Symbiosis is great for exploring modular composition and object-based composition. For example, the macroscopic musical structures created by our musical beta sheets can be thought of as large musical objects that interact with each other in different ways that I briefly mentioned above. One thing I like to do is to align these "sheets" so they contrapuntally target notes near structural amino acids like Histidine, Proline, or Glycine triggering their associated musical events (tempo changes, instrument changes, or articulation changes). That way these large textrual changes are triggered by the music itself and they don't just come out of nowhere, preserving a sense of drama and causality throughout the piece. I do this in a lot of my music, and an upcoming blog post will talk about this in the context of my recent work Syadvada.
I'll be excited to get this work performed when it's finished along with the other five movements of Cicada! This will likely be the most academically involved movement of the work, although I plan to explore modular composition even more with the sixth movement which is shaping up to be a kind of "choose your own adventure" style work. We shall see! Either way, if you made it this far thanks so much for reading and feel free to comment here or on my Facebook Page with any questions or thoughts. Hopefully you found this as stimulating as I did!