WEBVTT

00:00.000 --> 00:15.600
So, lightning talks are always interesting when a new team runs them. And I think we've done

00:15.600 --> 00:23.320
this one pretty well, but of course the proof will be in the pudding. And firstly, before we

00:24.160 --> 00:31.440
OK, I'm just seeing what's going on here. I should really be filling in. The caffeine hasn't

00:31.440 --> 00:39.920
quite kicked in yet to allow me to do that, so I'll keep sort of babbling a bit. So firstly,

00:39.920 --> 00:48.160
after this, we are going to be heading to Brutog, I think, which is near Central. So you're

00:48.240 --> 00:54.240
very welcome to join us. How long we're taking order to get there? I'm not sure. There

00:54.240 --> 01:00.280
may be some variants along the way, but if you want to carry on discussions or if you want

01:00.280 --> 01:05.920
to meet us there, either way, hang about at the end and we will work it out. There isn't

01:05.920 --> 01:13.120
anything. There are no more sessions either. We're going on after. So as mentioned

01:13.120 --> 01:19.280
before, some of us will maybe go on from Brutog to Bite Night. Some of us won't. The Bite

01:19.280 --> 01:24.720
Night is held at the Brussels Hackspace, which apparently is a pretty impressive place

01:24.720 --> 01:31.440
to visit. So if you've never been to a Hackspace before, who's come here for the first time

01:31.440 --> 01:43.680
for us, then, wow. So this is only my second time at Fosthen. The trick is to just go and

01:43.680 --> 01:49.040
talk to everybody. There's a thing about not getting tired, you don't have to sit and

01:49.040 --> 01:57.160
you've been amazing. We've been for pretty much the whole of the afternoon. And I think

01:57.240 --> 02:09.480
that's kind of unusual actually. Yeah, so, okay, guys, how are we doing? How much time are

02:09.480 --> 02:17.080
we giving each speaker? Three minutes. This is like slow lightning, but it's still lightning,

02:17.080 --> 02:25.240
actually three minutes. So you will see messages appear at magical in front of your face.

02:25.320 --> 02:35.080
Okay, so the trick is going to be, hold the mic. If people look like they can't hold

02:35.080 --> 02:39.400
you at the back, you should say, we can't hear you. So it's all going to be very quickly.

02:40.680 --> 02:49.080
I will try to do it quickly. It's not easy. So I'm small. I'm working in a university

02:50.040 --> 03:00.040
and I will speak about geeks and workflow management and so on. So the question is scientific

03:00.040 --> 03:06.680
production is about long-term trust. A long-term trust is about verification, be able to share

03:06.680 --> 03:12.280
and so on. So open source is one mean, but it's not the only mean. Another mean is to have long-term

03:12.280 --> 03:20.280
reproducibility. And the question is, how do we have a long-term reproducibility about workflow?

03:20.280 --> 03:26.760
So what is a workflow is a second of computation, but in real life is more than just a couple of

03:26.760 --> 03:35.320
steps. It's a concurrent, a multi-output, multi-output and so on and so on. So how do we have long-term

03:36.280 --> 03:40.920
reproducibility workflow? So there is two questions and usually people are taking these two

03:40.920 --> 03:47.560
questions completely orthogonal. The first question is, what is the computation environment

03:47.560 --> 03:53.400
inside each box? So where do I come from? The binaries, how are they produced and so on? And the second

03:53.400 --> 04:01.240
question is, how this language, how this graph is described? So which language and so on? So how

04:01.240 --> 04:07.720
you know? I mean, maybe you guess that I use geeks for the binaries. So what is geeks? I mean,

04:07.720 --> 04:13.320
it's in for your homework. Geeks is a package manager on steroid. So it's able to produce

04:13.320 --> 04:18.600
reproducible computational environments. So check out this webpage where there is a lot of resources

04:18.600 --> 04:24.840
and peerostokes in first them. So binaries is, okay, so the question is, is not about the binaries.

04:24.840 --> 04:32.120
The question is, how can we, can we plug the description of the graph with geeks? So the idea is

04:32.120 --> 04:39.000
the geeks workflow language is a really old idea by peerotero prints and rale. And the idea is to say,

04:39.000 --> 04:44.680
okay, package management is about graph management. Workflow is also a graph management. So can we

04:44.680 --> 04:50.840
build a workflow engine of the top on the top of the package management system? And it is, what is

04:51.000 --> 04:57.560
a geeks workflow language? It's a workflow domain specific language built on the top of geeks

04:57.560 --> 05:06.280
used as a as a library. So maybe you don't know. So yeah, I guess never heard about it. So yeah,

05:06.280 --> 05:11.560
it's not adopted at all because first it's yet another language describes the graph and so on.

05:11.560 --> 05:17.960
And it's based on this old kind of closing parenthesis, closing parenthesis language. And it's

05:18.040 --> 05:22.920
a competitive market. Like, you know, you have snake mega likes you next floor and so on. And

05:22.920 --> 05:29.640
maybe connecting both the description of of the workflow and the complexion of everyone was too

05:29.640 --> 05:37.800
descriptive. So check out Revenant. Revenant is a yet another implementation,

05:37.800 --> 05:41.480
powered by geeks. So we have a standard specification geeks workflow.

05:42.280 --> 05:48.520
Join us with Revenant and time's up. So here we are. Time's up.

05:52.520 --> 05:59.720
Please, please ping me on on email, uh, Master Don and so on. Let's share your long

06:00.360 --> 06:06.760
story about long-term Roblesibia workflow. Thank you. Thank you.

06:11.000 --> 06:17.320
Let's go. Okay. So I want to talk to you about a molecular visualization software called VTX.

06:17.320 --> 06:21.080
So why would we need those? Well, there is biologists that needs to seek what's important in

06:21.080 --> 06:27.160
directions. Sometimes for me, this micro molecular structure we don't see very well. And sometimes

06:27.160 --> 06:32.760
there is like a computer chemist that want to see just small molecules interacting with, uh,

06:32.760 --> 06:37.240
with proteins. So they have already, uh, some kind of tools like bi,

06:37.240 --> 06:41.880
bi, mole, chimera, and the MD. So why another one? Well, when we take, uh, like a hundred

06:41.880 --> 06:48.680
million martini beads from a world cell model, uh, it just doesn't work. So we had to do something

06:48.680 --> 06:54.200
better. Except maybe for VMD, which was kind of opening the stuff, but it was very hard to

06:54.200 --> 07:00.920
interact with the system as it was just only one, uh, point four, a frame per second. It's kind of,

07:01.640 --> 07:08.200
difficulties like a slideshow, you know. But that's where VTX shiny. It's, um, it could read like,

07:08.600 --> 07:15.720
this system and navigate it with 12 frame per second, which was quite nice. But what does it look

07:15.720 --> 07:20.840
like? So on the left, you have, uh, the image, the picture, uh, taken with VTX and on the right,

07:20.920 --> 07:25.240
it's VMD. So you, you, yeah, a picture, a big thousand words. And that's,

07:25.240 --> 07:30.920
because I only have three minutes. Um, and the upcoming version will feature a new,

07:30.920 --> 07:36.360
wrong-breaking way to see, um, protein surface. So currently, um,

07:36.360 --> 07:40.840
chimeraics and VMD and all that stuff, they use meshes to display, uh,

07:40.840 --> 07:45.640
molecular surface. So if you, uh, take a closer look, you see straight lines,

07:45.720 --> 07:51.160
trying to form round shapes, which is quite a satisfying to look at. When we, uh,

07:51.160 --> 07:56.840
the, the next version will feature, it will implicit like meshless, uh, protein surface,

07:56.840 --> 08:03.160
like we use, uh, gemage recalculation to render pixel perfect, uh, resolution of all the shapes

08:03.160 --> 08:09.960
of a proteins. Um, and it's quite fast. And, uh, it's main strength, uh, lies in it's, uh,

08:10.120 --> 08:16.040
lightweight. Um, it's in memory, uh, I mean, because RAM is quite expensive this day. And,

08:16.040 --> 08:23.240
uh, if we reduce the memory footprint, we could, uh, we could, um, display a way bigger,

08:23.240 --> 08:30.120
uh, more, um, systems and before up to, uh, two million items for this, uh, for, uh,

08:30.120 --> 08:38.920
eight gigabytes, uh, video card. Oh, that's fast. So, take away, it's already available. You can

08:39.000 --> 08:43.800
do a little bit, uh, following this QR code, uh, it's open source, obviously. And, uh, yeah,

08:43.800 --> 08:54.200
it's designed around, uh, performance and, uh, usability. Thank you, thank you, thank you, thank you.

08:55.320 --> 08:57.880
Oh, okay. Sorry. Sorry.

09:00.600 --> 09:01.640
Oh, sorry. Okay.

09:02.040 --> 09:03.640
Oh, okay.

09:08.120 --> 09:12.680
Play, everyone. My name is Fabian. I'm a PhD in a research assistant in Timishora, Romania.

09:13.560 --> 09:18.360
Alongside my colleagues and our supervisor, we wanted to create a web application for an

09:18.360 --> 09:23.560
ecological clinic. But in order to do this, we hit different problems. One of the problems was,

09:23.560 --> 09:29.240
okay, decont data, which is more important there. They want to see different modules or something like that.

09:29.240 --> 09:33.960
Okay, this coming into the, we have to make it 3D. And, after we want to see an evolution in time.

09:33.960 --> 09:40.680
So, go with 4D. In order to do that, we created, um, we found a way to go from 2D to 3D.

09:40.680 --> 09:48.040
Afterwards, we just added the slides and the way of getting it in time. For the discharge letters,

09:48.040 --> 09:52.600
which usually have the data regarding the, uh, blood samples and everything else,

09:53.480 --> 09:57.960
there was a little bit more complicated. And on top of this, we also wanted to automate

09:58.040 --> 10:01.720
kind of automate it and edit our all-based access. So, we can split it between

10:02.520 --> 10:06.840
radiologists, oncologists, and the front desk. So, the patient just go with the front desk. And,

10:06.840 --> 10:11.320
afterwards, everything is kind of automated. And now, I'm going to just go with each of them.

10:11.320 --> 10:15.000
For the natural visualization, most of it is based on precision, and this in toolbox, which

10:15.960 --> 10:21.560
helped us a lot for getting the data from 2D to 3D via the NRD conversion. And, after all,

10:21.560 --> 10:26.360
we just used PyroDomics to get different characteristics like not your volume, not your

10:27.000 --> 10:30.600
area. We also are trying to get the fractal dimension. We are still working on something,

10:30.600 --> 10:37.080
regarding that in, and you want to see if that's related to more into the oncology research.

10:38.040 --> 10:42.200
For the data visualization, here was a little bit complicated, because the discharge letters

10:42.200 --> 10:47.480
don't have a pattern. They are just written by hand in the word, even the tables are done with tabs,

10:47.480 --> 10:51.880
and it's very hard to get all the data. But we still managed to get it. We have an evolution in

10:51.880 --> 10:56.840
time for a graph or even a table. And in order to do that, we created an ensemble method,

10:56.840 --> 11:01.880
because first our deterministic model wasn't good enough. So, we kind of screwed up. But,

11:01.880 --> 11:06.920
with an ensemble method, we managed to get two, two, I models that work pretty good, combine

11:06.920 --> 11:12.600
with our DNA estimator. We have a voting algorithm, and the best output is selected, and added to the,

11:14.360 --> 11:20.600
database, in order to show it for the users. The database is among, most of the UI is done in

11:20.600 --> 11:25.880
React, and for the, and already conversion, and everything regarding the visualization,

11:25.880 --> 11:30.200
everything is done in a flash cap, in Python. And that's it. Thank you.

11:35.080 --> 11:36.920
I hope it is in time.

11:37.320 --> 11:49.320
So, yeah, I've been doing my PhD research and discipline students, my

11:49.320 --> 11:53.400
degree of study in past life. I'm a software hardware engineer, I come to the scientists,

11:53.400 --> 11:59.240
and the problem I'm trying to solve is as a hobbyist as well. I like working on the HDD SSDs.

11:59.240 --> 12:04.440
So, you can see the problem statement. Data is growing. So, currently I'm also working as an

12:04.440 --> 12:09.400
engineer, but I have a lot of synthetic data as well. So, what I'm trying to, with this

12:09.400 --> 12:17.400
lighting talk, trying to target is that, you know, as a computer scientist, I started working

12:17.400 --> 12:22.440
on the DNA storage, but the problem is the DNA storage has, does not have resources,

12:23.800 --> 12:27.400
accumulated resources for computer scientists. So, what I started, what doing was that I

12:28.280 --> 12:35.640
started collecting the open source DNA storage simulators, projects, so that, you know,

12:35.640 --> 12:41.400
I can make the life of the software hardware engineers, easier, who wants to basically proceed

12:41.400 --> 12:46.840
with this open source project. So, this is the seven steps of DNA storage. You have a source

12:46.840 --> 12:52.120
file where you basically have a binary format. You can work that into DNA sequences, then you do

12:52.120 --> 13:00.520
synthesis, and as, you know, in the synthesis, the DNA is basically broken down into sequences.

13:00.520 --> 13:04.200
Though, you have to cluster, you have to use clustering algorithms, so these are the seven steps

13:04.200 --> 13:09.720
of DNA storage. This is the report I'm basically trying to maintain now for the engineers,

13:09.720 --> 13:15.080
mostly working on the storage side, and this is something, you know, so-called awesome

13:15.080 --> 13:20.520
DNA storage get-up. So, basically, a curated list of all that projects, which was missing

13:20.680 --> 13:26.040
in the community. So, just please follow that, or if you have anything to fall to find in this

13:26.040 --> 13:31.480
report, please do that. And if you have any time tomorrow, tomorrow is a buff, which is

13:31.480 --> 13:36.920
photonics optics, so please drop by, we can discuss also something else storage. Thank you.

13:36.920 --> 13:44.920
Yeah.

13:52.120 --> 13:59.800
Okay, I'm the last one, and for our meeting, so, I'm Peyton. I'm

13:59.880 --> 14:06.760
a lecturer in health, and have some other science, but I'm here to present my work,

14:06.760 --> 14:13.960
observe, can I? So, in biology, isn't about if something happened, is about when and why

14:13.960 --> 14:22.280
where happened, something else. So, but in the risk comes from the continuous variables,

14:23.480 --> 14:29.320
the two is often forced us to pretend biology is binary, which is like in here,

14:29.880 --> 14:36.120
you can see just have high and low, but in the data here, you can see the gradients, which is

14:36.120 --> 14:41.000
a hidden messages, and also this is some of our researchers, you can see each live represent,

14:41.000 --> 14:49.320
maybe in patient, there will be a die, and the right will be alive. Okay, so, because there's

14:49.320 --> 14:56.120
limitations to think about categories, and just only high and low in the software,

14:56.200 --> 15:01.800
existing in our package at the moment. So, I developed the package, which is able to categorize

15:01.800 --> 15:06.920
in different and more than one. So, there's a free steps. On the first steps, there will be

15:06.920 --> 15:14.120
use an information criteria, a, i, c, or a, i, i, c, or b, i, c, and c, two, think about.

15:14.120 --> 15:22.520
Thus, the data is self suitable for four groups, or two is enough. And then, if that's two,

15:22.520 --> 15:29.880
even enough, or four enough, there will be use a genetic algorithms to more precise to know

15:29.880 --> 15:37.800
what the exact locations of the data. So, there will be, I avoid the problems of non-linear

15:37.800 --> 15:45.000
or use shape data. Okay, so once we note the cut off, of course, we need to do the

15:45.000 --> 15:54.520
validations, because in biology, having a risk for unknown would be a, there's also not,

15:54.520 --> 15:58.920
in a, sorry, about that. It's not in the data. So, which is why I needed to do the

15:58.920 --> 16:04.360
real and boosted up with sample links. So, in here, and the backgrounds of the three different

16:04.360 --> 16:12.360
pits, if we're presenting, so we're repeating 200 times, and you can see the cut off from the

16:12.360 --> 16:19.320
previous steps, because verify closely to the pits. So, which is the free step process for the

16:19.320 --> 16:28.840
project, a free step project for the package. So, right now, the package is in the, and is under

16:28.840 --> 16:33.880
reveal, in officer of cut out, if you would like to try, give a try, please scan in this

16:33.880 --> 16:44.360
Calcoc here, and thank you so much.

16:44.360 --> 16:50.200
That was the last one, right? Yes, yeah. I will let it open the fire close and work, it's, is it different

16:50.200 --> 16:57.240
to what was on this point? Because I could just plug in here. Yeah, okay. So, the last laptop switch

16:57.320 --> 17:02.600
of the day, it's a Mac, let's see what happens.

17:02.600 --> 17:14.280
Nope, thankfully. Well, yeah, we'll see. What does it do?

17:14.280 --> 17:32.280
Okay. So, oh, is it doing that as a, uh, good grief. How many different ways can they break?

17:34.840 --> 17:37.400
It's start mirroring, that's what I needed. There we go.

17:37.400 --> 17:55.320
Okay, right. Wow. So, um, yeah, I, I, yeah. This is it. This is it.

17:55.320 --> 18:04.440
Yeah, we do. Yeah, no, no, it will be. Was it fun?

18:07.560 --> 18:13.080
That was kind of a, yeah, we're smiling. We're smiling.

18:15.080 --> 18:21.240
I have to thank you both for being so freaking amazing just sitting there and being, like, timing

18:21.240 --> 18:27.800
demons, stream demons. We should have, like, rotated a bit more, I think that's something I would

18:27.800 --> 18:32.760
like, but no, no, I was trying, you know. I just want to sit at the back, you know.

18:33.720 --> 18:43.880
Ask or questions. So, we have a way for you to tell us whether you had fun, whether you

18:43.880 --> 18:53.080
didn't have fun, scan the QR code, go to our effort pad, make some notes, talk to us afterwards

18:53.080 --> 18:58.360
as well, because I was just saying before, actually, I don't have the pin for, but we will be posting

18:58.440 --> 19:04.520
where we'll be heading to the Brue Log, sorry, the Brue Dog in central, um, in the matrix trap.

19:04.520 --> 19:10.120
So, we'll be going, we'll put it in both, but one of the complicated things about phosdom is that

19:10.120 --> 19:17.080
because you have these phosdom rooms for the devrooms, um, but they kind of disappear. So, we have our

19:17.080 --> 19:22.520
permanent space, which you can also join. It really doesn't want to join. There are things

19:22.520 --> 19:26.680
that you might want to ask. So,