First time in more than a year met in three-dimensional space with a researcher. Ms. Krithika Arumugam from  Singapore Centre for Environmental Life Sciences Engineering. She helped me understand DNA Sequencing and genomes.

Krithika Arumugam SCELSE page: https://www.scelse.sg/People/Detail/770a9e99-97c2-4c13-b470-a1631db35409

Link to the work discussed in the podcast:https://www.nature.com/articles/s41522-021-00196-6

Subscribe to the podcast: https://www.lefterisasks.com/listen

Subscribe to the weekly newsletter: lefterisasks.com/newsletter

Buy me a coffee: https://www.buymeacoffee.com/LefterisAsks

Join the Facebook Group: https://www.facebook.com/groups/544815233347328/

If you're on a sampling from an environment, there's going to be

like, uh, hundreds of bacterias in it.

So just imagine, uh, there's going to be like millions of jigsaw pieces and

you don't know the original picture.

Uh, it's going to be very tricky to sort of assemble them and picture

what's exactly there in the community.

Greetings my good humans and welcome to Lefteris asks science edition.

Number 25.

I am.

Lefteris the annoying guy that calls academics and scientists and ask them

questions until I understand what, how and why they do what they do this week.

For the first time in more than a year, I met a person in three dimensional space.

I traveled all the way to Nanyang technological university and met Ms.

Krithika Arumugam from the Singapore center for environmental life

sciences and engineering, and she helped me understand what genomes

are and what DNA sequencing is before we go on with the show as always.

We have some housekeeping.

If you enjoy the show, please subscribe to it and share it with

your friends that might like it too.

Follow me on Twitter at Lefteris_asks.

And we also now have an Instagram page under the same

name, which is very exciting.

Also, I just made a new Facebook group called Lefteris , ask science,

come join the group to ask questions, find out about updates and much more.

Lastly, I have a weekly newsletter where I share my favorite news from the

world of science and academia have small explanations and links to the research

for anyone who wants to find out more.

If you like that, go to the show notes and click that link

to subscribe to the newsletter.

Lastly, in the show notes, you'll find links that you can support

me in doing this by donating.

Let's now meet Ms.

Arumugam.

I'm Krithika Arumugam.

I'm a, bioinformatician at the Singapore center for environmental

life science engineering, which is a research center of excellence

located at Nanyang technological university here in Singapore.

So, um, I finished my undergraduate , computer science

and engineering from India.

And then, um, I moved through my master's here in Singapore and

bioinformatics, or just to be at the center of an Interdisciplinary field

do not adjust your podcast sets.

I did say in the beginning of the show, we're going to talk

about genomes and DNA sequencing.

Ms.

Arumugam moved from computer science background to bioinformatics.

If you want to be a part of multidisciplinary study, then you'll

probably have to juggle a lot of different information about different fields.

How has that experience for Ms.

Arumugam?

Yeah, I would say it was challenging initially.

Um, well, um, I was completely new to all the biology constantly.

I had some modules in biology in my high school.

That was my last, you know, studying biology, but then after coming, yeah,

like, uh, yeah, used to be, it was a bit challenging jumping into a new topic.

Uh, you'll get to know a lot of things.

It was quite interesting.

That's right.

But, uh, like using your informatics skills to understand, you know, complex

things and biology was very exciting.

Yeah.

It was always exciting for me to see unfamiliar terms, even from

the title of the paper, while I might have seen the word genome before the

word replicon did not sound familiar.

Lucky enough.

I have someone that works with both genomes and Replicons to

explain to me what is, what.

So, I guess, uh, do you know, must be a very familiar term,

if not, it's, uh, it's basically a genetic material found in any living organisms,

like, uh, so which is actually the DNA, the DNA, Deoxyribonucleic acid, I guess.

Uh, you know, there is a sort of, uh, uh, helix bridge in Singapore,

near the Marina Bay sands.

So that's a DNA inspired structure.

So that's how a DNA look like it looks like.

And, uh, So once you decode that or for any particular organism, then,

uh, that's actually gone the genomes.

So, uh, which is actually responsible for, uh, all the functional aspects

of living organism or how actually a living organism looks in case of humans.

Like how you look, uh, what's your hair colour.

So everything is encoded in the DNA.

Okay.

So the genome consist of, you know, the chromosome, which is

the primary genetic material.

Um, and then data other, uh, genetic material as well, which

we call them as non chromosomal replicons which are smaller in

size compared to the chromosome, but, uh, they are still useful.

And, uh, they have other different function aspects compared to the main

chromosome, make better the primary color.

As we learned, genomes are the instructions and rules of an organism

and follows to grow and develop.

Your DNA or the RNA if you are a virus, one thing you might have heard about

when it comes to DNA and its genomes is a term called sequencing, DNA consists

of four small, different compounds, cytosine, guanine adenine, and thymine

the long helical structure of a DNA consists of different

sequences of these four components.

So sequencing is finding out what is the structure of the DNA when

it comes to these four components?

Sequencing is basically the process of,

uh, reading the DNA and coding.

What's actually in the DNA.

So that's a technical term.

We call it sequencing.

What people normally do is, uh, they collect samples or, uh, If it's a single

bacteria, they tried to grow it in the lab and then, uh, sequence it as in

sequence it, as in meaning, uh, decoding, what's actually present at the DNA to

understand the function of that particular organism in this case, bacteria.

So there are different methods, which does it.

So once you collect the sample, you extract the DNA out of it.

So this is all done in the lab.

So the DNA is actually sheared the processes.

You start with shearing the DNA.

That's you cut down the DNA into multiple different lengths and then the machine,

tries to read, uh, each of the fragments.

So.

And the machine actually gives you an output in the form of text file.

So, okay.

Text file.

As in, you cannot open it with a normal text editor, it's going to be quite big

it's in terms of giggle, it can be as big as gigabytes or terabytes as friendly.

So, I mean, that's when we step in the computation people step in because as

the data gets bigger, you need like specialized people with different

expertise to look at the data.

Right.

So, okay.

The actually the file.

It's actually,a text file, uh, made up off, uh, different characters encoded

in the DNA, which is actually translates to the compote and coding and the DNA.

So the DNA is actually made up of, uh, four compounds.

We call it, uh, A T G and C, which is actually Adenine,

thymine gyanine and cytosine.

So.

Theshort form is ATGC.

So you have a text file with, you know, permutations and combinations

of ATGC, which is actually the DNA being encoded and decoded.

Uh, uh, so since you have a sheared DNA fragment before

into multiple fragments, so.

The land of the DNA, fragment is quite small, right.

But the original size of the DNA of the chromosome and the living

organism say, for example, microbe, it's going to be like five megabase,

but, uh, the technology allows you to only share the DNA and then

read it out and smaller famines.

So, okay.

So since the fragment size is smaller, it becomes difficult to reconstruct the

original chromosome of the bacteria.

Uh, okay.

So an easier way to visualize this as people usually compare it to a

jigsaw puzzle, I'm sure you must have played it as a kid or, yeah.

So, so you.

Okay.

In a jigsaw puzzle, you don't know what the picture is, but

you have this picture is actually cut down into smaller fragments.

So you try and piece them together to form the bigger picture, the bigger picture.

Right?

So, okay.

In this case, it's a bit tricky because if the jigsaw pieces are going to be

smaller, there's going to be many pieces.

So it's going out.

If you don't know the original picture, it's going to be difficult to, you

know, uh, put those pieces together and reconstruct the original picture.

So in this case, we, uh, uh, you know, uh, jarring the small genome fragments

that can be like millions of fragments.

So putting them together and trying to find the original

chromosome is always challenging.

Right.

So even if it's challenging, even if it's a single bacteria.

So for example, if you're sampling from an environment, there's going to

be like a hundreds of bacteria in it.

So just imagine, uh, there's going to be like millions of jigsaw pieces and

you don't know the original picture.

Uh, it's going to be very tricky to sort of assemble them and picture

what's exactly there, uh, community.

I really love that.

Puzzles example.

So when scientists want to find out, for example, what types of

organisms live in the Lake to take a sample and try to piece all of the

fragments together to figure out what is the complete picture they make?

It's like, if we take.

10 1000 piece jigsaw puzzles through all of the pieces in the same box and

try to create the 10 different images.

It would take a lot of effort and time to achieve that.

Now, one thing that would make things easier would be if the pieces

were actually bigger and here is where the terms short read and

long read sequencing will be used.

Technology has advanced, uh, to an extent that, uh, we

can increase the size of the pieces as in the DNA fragments, which are being read.

So normally what we do is we do short read sequencing.

So, uh, in short read sequencing, um, uh, the error rate is very negligible.

So, uh, today it may still use Sharpied sequencing, but, uh, one of

the limitation is that the lent of the read or lent of the DNA fragment,

we call it a read in technical dorms.

So the lens of the DNA fragment, it can read us.

It can go up to 300 base pair or 300 characters if you could call it that way.

So the technical term is being spare.

So each of the is a base and T is a base.

So you can read up to 300 base pairs.

Okay.

Uh, the process of, uh, trying and putting the genome fragments

together is going to be tricky.

So it's.

It becomes easier if the genome Frackman, you know, it can be, uh, if it's a

bit longer than the, the sequencing machine, you know, can, uh, read the

DNA, fragment, uh, up to a bigger land.

It's going to be easier comparatively.

When you have a bigger jigsaw piece, you know, it's going to get easier, right?

So that's what the long lead sequencing here actually means.

Now.

How do they actually do the assembly of the genome and how do they know

that they got the correct picture?

So, what does is, uh, it takes the beads and dry and

bothers them to produce a quantity of this, uh, section of the news.

So it doesn't overlap.

The reads are going to be marched.

So like I said, the bead is composed of ATGC.

Right?

So if there's an overlap in another fragment, those two are

going to be marched and we try and produce the algorithm twice and

produce extended fragments to try and make the reads a bit longer.

So, so once, uh, and a lot of them has, uh, processed the reeds.

What we get out of it is called conflicts, uh, which are actually

extended fragment of the leads.

So it's just a different terminology.

Sure.

Uh, so we call them con things because they are contiguous

sequences, I guess it's a sharp font.

Yeah.

So.

Okay.

and then there are different techniques, uh, uh, to try and combine the context

based on the, uh, uh, based on the characteristics of the conflicts.

As in, um, uh, how, uh, uh, there are different characteristics of the Cod

things like you can take into account the abundance of the context, like how many

leads were used to make that context.

So of the more, the number, uh, the reliable the content is going to be.

And then we can group context together with, uh, similar

abundances, assuming that they are coming from the same bacteria.

I should.

Okay.

So, I mean, this is one kind of characteristics.

There are multiple characteristics as well.

So sometimes we combine all those characteristic to try and group those

contexts together to see, uh, you know, uh, which of these contexts belong

to, or, uh, come from which bacteria.

So, okay.

In this case, uh, Uh, the back, the genome is still made up of multiple contexts.

It's not one complete content.

So that's what we are trying to achieve with long lead sequencing.

We are trying to achieve, if we can acquit one single content, like, uh,

continuous, uh, content, uh, of say five megabase band and blend, which

is approximately the size of a back.

Yeah.

So in case of short leads, you still can get fine obvious pair, but they're

going to be fragments of context, which is going to sum up we'll fight.

So they can be like a hundred KV, a hundred KB or NMB.

So everything together.

So they are instilled in multiple fragments, uh, because there are,

uh, Since they are in multiple fragments, uh, uh, we might not

know the ordering of the con things.

You know, if a particular content is going to, you know, be located, located

at position one, and it's going to, uh, be difficult to please do this context

together or the order in which they.

Actually come from.

Yeah.

So, so those are the limitations associated with charter and

sequencing and it can be difficult to sequence or, uh, plays the comp.

There can be complex regions in the genome of any organisms.

So.

If the lead to shorter, we might not have sequenced those complex regions.

So it's going to be difficult to piece them together.

So that's why we use long read sequencing, uh, to actually, um, check

if we can, uh, uh, get a continuous.

Genome sequence instead of multiple fragments.

Yeah.

So that's why we were able to actually, uh, uh, see in this paper, like,

uh, we did be bad able to extract around grade two genomes, which was,

um, like complete tools to genomes.

Okay.

Meaning it's in a single fragment, not fragment, meaning it's as

soon as single sheepish seats.

Right.

The benefits of actually having a face to face interview.

I was able to see both the machines that were doing the sequencing, but

also most importantly, themselves, it is astonishing to see a text file that

is gigabytes in size, in my complete ignorance of homicide among and works.

I asked if life would be simpler for her.

If instead of sequences of letters, she would find a different

way to visualize the data.

So it's difficult to wish you advise

the reader at the bead level.

Yeah, but we, uh, well, uh, at the level, but, um, there are different ways you

can, depending upon what do you want to work looking forward in the data,

depending upon your research Westin.

So if you want to look at.

Uh, the taxonomy taxonomy to the content often species and the data

then, uh, uh, we try and map the leads to, uh, existing databases.

So, uh, if, uh, you know, how many leads mapped or a selected species or

a certain gene is off the back the app.

And if the number of leads mapping to a certain bacteria is more than making.

You know, say a certain, certain percentage of bacteria

is found in that sample.

So something like that, but, um, uh, that, that's how we were

analyzing the data initially.

But then as the, uh, you know, uh, assembly and gardens started

developing it's, uh, it becomes much more reliable when you try and piece

those things together instead of just mapping the global database.

We try and reconstruct or we didn't cheat.

So that gets more interesting.

And we can actually know what kind of bacteria is actually in

there and what are they doing?

Yeah, yeah, yeah.

The functions of it.

So if you try and like where the genome, then it's easier to understand

how that particular bacteria work,

uh, how it's responsible for certain things, certain

processes and the sort of stuff.

Puzzle solving in this magnitude doesn't happen on a local computer level.

These algorithms require a lot of computational power in order to give

results in a relatively short time.

But even then the time is not as short as you think.

Sequencing promising sequencing machine depends

upon, uh, the throughput of the data.

It can take a day or two, so to sequence it, but when you're processing it,

processing the data, uh, it depends on.

What do you actually want to do?

So if you're doing a taxonomy analysis, uh, uh, it can take a few

days, uh, but if you're doing an assembly and garden, so, uh, I mean,

there's also a sort of a limitation to the existing assembly algorithm.

So, uh, so the data size keeps increasing, but, uh, you know, it's difficult to

catch it computationally with the devil.

As well.

So, so for example, we had been trying to assemble, uh, um, uh, I can't remember

the exact number, but maybe it on the.

A billion reads.

So, um, with the existing capacity computeration capacity, we have

the metagenome assembly, uh, uh, no two or three months, I guess.

So, I mean, it doesn't make sense to wait for that long.

So instead we try and.

Sort of compress the data or subsample it randomly in a way that, you know,

you could answer your questions sooner.

Yeah.

So, yeah.

So it depends on what you actually are looking for or what kind of

questions you're looking to answer.

So.

Uh, so usually, uh, the, in general, we, if you're generating say around,

uh, one run of Hi-C Hi-C is the sequencing mission, the Catan of

sequencing mission it's mostly sheltering sequencing is mostly done in

Illumina Illumina is the company name.

So the kind of mesh we use is high seek.

So high seek generates around one round, of high seekgenerates around, um, uh,

600, approximately 600 million reads

so if your community is going to be complex, that is if.

You think there's going to be like conduct of bacterias or

hundreds of microbes in it.

You have to sequence more.

Yeah.

Only then, you know, the sequencing depth has to be high.

You don't need that.

And you know, what kind of microbes out in there.

And if, uh, if doc under certain microbes are.

Less than the community.

It's going to be difficult if you sequence, if the sequencing depth is low,

it's going to be difficult to recover genomes of microbes of lower abundance.

So the higher, the sequencing depth better your chances of recovery.

Yeah.

So

basically, right.

So yeah, so for, uh, For example, one run of high seek generates

around 600 million reads.

So I, uh, yeah, so, so you do initiate quality checks as well of the raw data.

And then if you're doing a mega genome assembly, um, we

usually, uh, split the data.

I mean, so for example, you can, okay.

So.

Uh, they can be multiple samplings sequenced in one drum.

So each of those, uh, sound booths might have like say, uh, if there

are 10 samples, then they might be around 60 million reads per sample.

So you can assemble them sample wise as well.

So that's going to be much positive in terms of things.

Sure.

So if you're putting all the samples together and assembling

600 million reads, that's good.

Do you need more Ram?

That's going to take like a few days, a few weeks.

Sure.

So, and then once you get the results from the assembly, uh, So, I mean, if

it's shocking and sequencing, you're not going to get the complete sequence of

back together going to be in fragments.

Right.

Even though you assembled them.

So we use other techniques called my dogs, you know, bending.

Uh, that's how, uh, earlier, when I explained we try and group those contexts

together based on their characteristics.

So that process is actually gone by that, you know, being.

So there are other downstream analysis as well to evaluate, uh, if the bin or all

those contexts belonging to a particular genome particular bacteria, if they are

complete, we have to analyze that as well.

So there are other than downstream processing as well.

So yeah.

Yeah, I could see the one they wanted two months or something for yeah.

For wonder highest seat and then there, and then you can interpret, so, okay.

These are the kinds of bacterias in their store.

We can then check the functions of it, et cetera.

So it depends on your research question.

Yeah.

Time taken.

So the initial processing can take a month.

Yeah, yeah, yeah.

And that's it for another edition of Lefteris asks

science, DNA sequencing is a big puzzle solving exercise, which

sounds really, really exciting.

And imagine that sometimes if you have enough results of a sequence that

you can't match to anything in the database, you might discover a new kind

of species, which is always exciting.

I'd like to thankKrithika Arumugam for her time and the description of the

episode, and you'll find links for her bio and the work we were talking about.

And thank you for sticking around until the end in the show notes,

you will find ways that you can support me in doing this.

One is a way you can support me, especially just sharing

the episode with a friend.

I really appreciate it until we meet again, take care,