Introduction to Genomics for Engineers
learngenomics.dev
learngenomics.dev
Also gotta shout out to these incredible molecular animations by WEHI: https://www.youtube.com/watch?v=7Hk9jct2ozY
But they produce short reads, and because DNA is full of repetitive fragments, it's not always clear where the read came from.
We also have two copies of genes, which also further complicates matters.
The first startup where I worked, developed synthetic long reads on top of Illumina's hardware. We could stitch together 50kbp reads, which really helped with de-novo sequencing.
For example, sequencing instruments include base quality strings in the output. Base qualities are estimates how likely the instrument got each sequenced base right. But most people don't want to store that much noise, especially when the actual data is highly compressible. So the base qualities get quantized using more or less principled methods that seem to work well empirically.
Read aligners make similar estimates of how likely they got the correct alignment for each read. Those estimates are typically based on simplistic models and a number of assumptions. There are two main components in the estimate. One is based on comparing the primary alignment the aligner chose to the secondary alignments it also found. Another is an estimate that the aligner didn't find the correct alignment, because that part of the sequenced genome is too different from the reference. The latter is obviously handwavy. And the aligner cheats in the former. Because people don't want to wait 10x or 100x longer for better results, the aligner gives up early and estimates how good secondary alignments it might have found if it had actually done the work.
And then there is variant calling. At some point, the state-of-the-art callers were statistical. But then people got better results with neural networks. Or at least the results were empirically better.
Meanwhile human geneticist doesn't even need to collect any data. Reference genomes are always being improved. People dump their data into public repositories (at least public for other credentialed researchers). A couple emails and filling of approval forms and you too can have access to 10,000 patient samples of some human disease already sequenced for you to an acceptable depth. Of course you will still need to pay for downstream compute needs in money and your time crafting the analysis to suit your reasoning, but still, half the battle is already won when you work on these well trodden paths. So much necessary groundwork has been performed by others for you already.
I'm not a maths person; that's just what I've observed.
I'd strongly recommend in reading up on the parts of cell biology that come after this. Otherwise you'll get the wrong impression of how messy biology actually is.
There's the X and Y chromosomes, those produce a binary result (unless you have a genetic anomaly). And after that comes the messy and fuzzy parts I mentioned, where those genes trigger changes in hormone levels and development. And those parts are analog, very complex and contain a lot of different parts. So the outcome is not binary anymore.
There are more combinations than merely having only X, or an XY combination. And there is more fuzziness even in the Y and X expression, as you said. It's fuzzy all the way down. The tale of Binary results has always been from compression of reality: Always has been.
But you're right, the full range of biological possibilities is very fuzzy . SRY itself a just a regulatory switch that other sex-linked traits are conditionally dependent on. If the switch gets broken, you develop as female. If genes that support the switch break, you might develop as female. If a sex-linked trait downstream from SRY mutates, then pretty much anything can happen. And other species do sex determination completely differently. Hell, a lot of bacterial sex basically involves throwing pseudo-viruses at each other.
It's definitely possible to learn enough to be productive within a few months, but to actually comprehend and understand the underlying biology takes much, much longer. I still don't understand much of what is presented by people from other labs outside of my specialty.
The course was mentioned in a recent whoishiring thread. Sounds like it could be a purposeful place to work.
I'm not affiliated in any way, just found it interesting.
This is a weird description, because ... it is not really "broken up". Each chromosome could be shuffled and put into different cells in different numbers. Now, it is unlikely that the resulting cell would be viable or useful, but my contention here is the "broken up" part. Chromosomes are just a way to handle the genome set. There are reasons why bacteria do not have chromosomes and this has mostly to do with the amount of DNA. To call this "breaking up" is a very strange description. (Size is not the only reason; duplication of the DNA before cell division is another important factor; bacteria usually have just one origin of replication, eukaryotes have several on each chromosome, otherwise the S-phase in the cell cycle would simply take too long.)
> Each genome is a biochemical database that, if properly accessed, can inform how our bodies function.
This is also a very strange description, aka "biochemical database". Not everything in a genome has a role with regards to biochemistry or metabolism. Some is just regulatory RNA; some of this relates to metabolism, but you also have e. g. piwiRNA or silencers of transposons and so forth. That in itself has only very rarely a biochemical function, with some exceptions (e. g. I would classify tRNA as related to metabolism, and many viruses have tRNA or use tRNA as quick-starters, but most of those regulatory RNAs do not have any function for metabolism directly, other than e. g. repurposing energy towards their own reproduction).
To me it seems as if the article was written by an engineer. That's fine, but it also means that the thinking is quite biased. Genetics is not quite so easy to engineer; a good example are leaky promoters used in synthetic biology (just ask the people who use such promoters how to make them un-leaky) or off-target cleavage effects in CRISPR-Cas(9 or whatever is used); I am pretty certain they'll give excuses as to why 100% accurate gene therapy isn't yet ready for the masses. And they'll do that for quite some years to come, I bet, usually hiding behind "it will cost too much" - when in reality, it should cost very little, if it were to work, rather than this just becoming the new meta-milking scheme.
The DNA of a nucleated (a.k.a. eukaryote) cell is indeed split into multiple chromosomes and the number of the chromosomes and the number of genes on each chromosomes and the sequence of the genes on each chromosome are normally constant for a species and very similar for closely related species. The cells that have an incorrect number of chromosomes (which happens when a cell division does not work correctly) will normally die soon, because their DNA is incomplete.
Only when a cell has all the normal chromosomes, but also some extra chromosomes, it has a chance to survive, even if the extra chromosomes may interfere with some internal processes. Because superfluous chromosomes are much less harmful than missing chromosomes, there have been cases, rare at animals, but frequent at plants, when the entire genome has been doubled, when some cell division has failed, but the descendants of that cell have survived.
There are animals for which the number of the chromosomes and the sequence of the genes on them has remained unchanged for many hundreds of millions of years, though there are also animals where the DNA has been completely rearranged, because some chromosomes have fused into a single chromosome, other chromosomes have split into multiple chromosomes, and on some chromosomes the genes have been shuffled.
Nowadays, it is known that it is very likely that the common ancestor of all animals except comb jellies had 29 chromosome pairs (humans have 23 pairs). In most animal branches there were more chromosome fusions than chromosome splittings, so in the present most animals have fewer than 29 chromosome pairs.
Among the animals with the most conservative genomes are some sponges, some jellyfish, many echinoderms and some of their relatives, i.e. acorn worms, the lancelets, some nemertean worms and some bivalves.
I had a software engineer do a bioinformatics PhD in my group several years ago, and I wish this had existed to show him. It would have saved all of us a lot of time, and him a lot of angst and confusion.
I used to think it would be the other way around, but having supervised grad students from both sides of the divide, I can say with some certainty that it's easier (in general) to teach molecular biology majors coding, than it is to teach software engineers cloning.
(There are exceptions of course, both ways.)
The reality is that to succeed in this field now, you need both sets of skills.
It doesn't read to me as if it's written by an engineer. It reads to me as if it's written for engineers. A well written piece of introduction material bridges the gap between it's audiences, and must leave out significant detail.
I've seen it mentioned by one of their people in a recent whoishiring thread and found it neat.
This looks like a great guide to read.
But I think before diving deeper and reading the rest of the guide, which granted it is from employees working in a lab inside of a hospital, I'd like to get the expert opinion of a geneticist or an expert biologist with years of experience in genomics to iron out any issues in the guide or give an additional proof-reading review.
These aren't things you get by reading full chapters of an introductory textbook. These are the things you get from reading a few chapter summaries. (Of even just a book's intro.)
1. Everything, and I mean everything, is stochastic. There is nothing in biology that is a guaranteed "if X then Y," there's nothing in biology that is a guaranteed "X is used for Y," "X is only used for Y," or "only X is used for Y." Even stuff that seems like it _should_ be that way isn't. RNA folds into useful shapes, the codon table varies between organisms, enzymes will target and modify multiple substrates, and metabolic pathways can and will run in both directions depending on the circumstances. Understand this, internalize this.
2. Biology is a physics problem, not an informatics problem. There's no API boundaries between different "layers," because everything is molecules jostling against other molecules. This means things like the geographic distribution of molecules within a cell can and will have serious effects on gene expression and biological processes. What's more, that "no API boundaries" extends to the cellular level - the "cell wall" is a thing cells use both sides of, bacteria regularly swap genes and genetic material, and metabolic pathways will pass between unrelated organisms.
Basically, everything we've ever done to turn engineering into a tractable problem does not exist in nature. Nature grabs whatever happens to be right at hand and shoves it into use. Consequently, it is _devilishly_ complicated to model, because every simplifying assumption you want to make has exceptions that stack to "your model works perfectly exactly once on a Tuesday at 3pm, but only if the humidity is over 72%." This is also why you'll notice your lab biologists are the kind of superstitious that would make a pagan soothsayer say "oh come on, it's not _that_ bad."
It's an awesome, amazing field, and there's huge contributions you can make as an engineer, but step one is shut the hell up and listen to the scientists, and step two is to learn that every time you hear "X does Y," your next questions should be "under what circumstances?" "how often?" "when doesn't it?" and "what happens then?"
For someone interested in the field with only a computer science background, are there any jobs in the field that could reasonably be transitioned into without any formal biology education?
Maybe a section on RNA degredation and DNA stability and how it would affect sequencing would be nice.
Also, down stream analyses are largely missing e.g. differential analysis, pathway enrichment. Not to mention newer single cell techniques and their up/down sides. But good start!
https://news.ycombinator.com/item?id=33734846
2022 (310 points, 82 comments)