KERNEL Issue 6, 2021 - IISc
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Research Newsletter of
the Indian Institute of Science Issue 6, 2021
KERNEL
Editorial MODELLING A PANDEMIC
Despite their limitations,
models and simulations are
useful for tracking the spread
of a pandemic like COVID-19.
In this issue of Kernel,
read more about such tools
developed by IISc researchers
to gain deeper insights into
disease transmission.
Aerosol cloud formed between two people after 70 seconds of conversation (Image courtesy: Sourabh S Diwan)
We also feature stories
PROJECTIONS AND SIMULATIONS HAVE HELPED PROVIDE A BETTER
on how scientists have
HANDLE ON UNDERSTANDING DISEASE SPREAD AND MANAGEMENT
visualised ferroelectricity in
nanomaterials and gauged In January 2020, India reported its first of disease spread. Factors such as the
how machine learning cases of COVID-19 caused by the now pathogen’s virulence, the number of
infamous SARS-CoV-2 virus. It was declared a people affected so far, the severity of
models match up to humans pandemic by the World Health Organisation infection, the mode of infection spread
in “seeing” objects. Finally, on 11 March 2020. A year and a half later, and so on are typically considered while
our lab feature focuses India is still struggling to escape from building a model. Policy makers have
the clutches of this deadly disease. With relied on such models to devise suitable
on research in flexible a potential third wave lurking around the interventions that can prevent or reduce
electronics that can pave corner, how do we prepare for what’s coming? dangerous outcomes.
the way for novel materials
In this quest for better preparedness, Many IISc researchers have also been
capable of bending and modelling and simulations continue to involved in developing such models,
folding. play an important role. These models are simulations, and analysis tools for
built based on the current understanding managing the response to COVID-19.
Continued on page 2
Continued from page 1 Modelling vaccination strategies (Image: Nihesh Rathod)
For example, during the first wave, a model and explains Diwan. He warns, however, that this model incorporates seroprevalence
dashboard that projected the requirement of such scenarios are more complex in real (the extent of population exposed to the
medical inventory like oxygen supply, masks, life, where factors like the exact phrases virus) and seroreversion (the extent of
PPE kits and ventilators was developed. used during speaking, cross ventilation, previously infected population which is
Other models studied the effectiveness of and ambient humidity can also have a direct now susceptible) while making decisions on
random testing, contact tracing, lockdown-like impact on disease spread. different vaccination strategies. It evaluates
restrictions, and isolation on the spread of the the vaccination requirement and proposes
disease. Another model that has proved useful three different roll-out plans, based on
in policymaking is a partial differential population size, seroprevalence, and number
One of these simulations looks at air flow equation (PDE)-based model developed of cases – including active or recently
patterns to understand how COVID-19 can by Sashikumar Ganesan, Professor at the recovered infections in a state or district.
spread from one person to another. Scientists Department of Computational and Data
have shown that COVID-19 is primarily Sciences (CDS) and Deepak Subramani, Their model shows, for example, that by
transmitted via droplets from an infected Assistant Professor at CDS. Their model increasing the number of vaccine doses
individual. A research group led by Sourabh S projects the potential number of infections to about 4 lakh per day in Karnataka
Diwan, Assistant Professor at the Department in the medium-term (>6 weeks) based (assuming an average of 1 lakh doses per
of Aerospace Engineering (in collaboration on the current caseload and multiple day as the baseline), by 31 August 2021, the
with International Centre for Theoretical scenarios. case load could be reduced by about 55%,
Sciences, Bengaluru and NORDITA, Sweden) when relatively strict non-pharmaceutical
has been studying the patterns of air ejected An important parameter it incorporates interventions (NPI), like lockdowns, are
when a person coughs, sneezes or even speaks, is the age distribution of the population, followed. On the other hand, when the NPIs
using insights from the analysis of cumulus specifically the age distribution of the are slightly relaxed, the case load could
cloud evolution. infected people, which largely determines reduce only by 37%, and when there are
the preparedness required from healthcare no interventions at all, by 13.5%. “[This
Over the past few months, the team has systems. It also includes parameters such model] can help policy makers determine the
performed simulations on the SahasraT (CRAY as rate of testing, the loss of protective level of non-pharmaceutical interventions
XC40) supercomputer to examine COVID-19 antibodies over time, and extent of needed to not overburden the healthcare
transmission through transport of aerosols, interaction between individuals. The model system and come up with good strategies to
which are tiny droplets that largely follow then uses this information to predict the prioritise vaccination across districts,” says
the air flow generated during speaking and disease spread and its severity. Crucially, Sundaresan.
can remain suspended in the air for hours the model predicts how cases will rise or
even after the speech flow has died out. For fall under different scenarios, such as no But with any model, there are several
instance, their simulations show that when lockdown, complete lockdown, night or challenges. A pandemic like COVID-19 is
two people of the same height are engaged weekend curfew in some areas and so on. laden with uncertainties, partly driven
in a short conversation, the jets coming out This capability is important for science- by new variants and mutants. Any model
of their mouths face mutual opposition after informed, data-driven policymaking. “The should try to include the many underlying
travelling a certain distance, and therefore only projections are good for the next 4-6 weeks, uncertainties at every step, points out
a small fraction of aerosols released by one and once every 4-6 weeks, we update the Vishwesha Guttal, Associate Professor at
person reaches the other person’s face. This parameters, introduce new scenarios and the Centre for Ecological Sciences. He also
considerably lowers the risk of infection. But if evaluate our model’s performance since the says that incorporating the most recent data
there is a certain height difference between the last update,” says Subramani. on disease spread, infectivity rates, target
two people, the speech jets can slide over each population, and so on, as well as genomic
other and reach the other person more easily, Ganesan adds that they have discussed and surveillance data – which could hint at the
and therefore carry a higher risk. shared their projections with authorities emergence of a new strain or variant – are
involved in COVID-19 management at both crucial.
Another way of decreasing risk of transmission, the local and national level.
the team found, is lateral separation, where The time frame is also important, he says.
two individuals are not directly face-to-face. Yet another recently developed model “One must avoid long-term projections
“[Tilting the] head away by about nine degrees provides suggestions on how to manage because, at the end of the day, there are
can ensure that even when you’re not wearing vaccination strategies. Led by Rajesh many aspects of infectious diseases that are
a mask, you are generally at a lower risk of Sundaresan from the Department of dependent on human behaviour, which in
infection than when you are looking directly Electrical Communication Engineering, in turn can be affected by policies.”
in each other’s faces, especially when you are collaboration with the Indian Statistical
separated by a distance shorter than six feet,” Institute and University of Virginia, - Sangeetha Devi Kumar
KERNEL | ISSUE 6, 2021 | Pg 2
Image: Pavan Nukala
VISUALISING OXYGEN MOVEMENT
IN FERROELECTRIC MATERIALS
STATE-OF-THE-ART MICROSCOPY HAS HELPED SCIENTISTS UNVEIL THE MECHANISM BEHIND AN UNUSUAL
PHENOMENON
A new study by an international team But in 2011, scientists showed that (as the device is made smaller), oxygen
of researchers, including those from hafnia-based oxides could exhibit conduction became more robust. These
IISc, reveals how an unusual form of ferroelectricity even when they are findings were also confirmed by X-ray
ferroelectricity arises in certain nano-sized nano-sized, and that this ferroelectricity diffraction studies carried out in Sweden.
materials. The team includes Pavan Nukala, in fact grows stronger as the material
Assistant Professor at the Centre for Nano size gets smaller, opening up numerous Oxygen migration occurs due to imperfections
Science and Engineering (CeNSE), IISc, opportunities for microelectronics. or “vacancies” in the crystal structure,
and former Marie Curie Research Fellow at However, it wasn’t clear how explains Nukala. “These structural defects are
the University of Groningen, Netherlands – ferroelectricity happens in these nano- the key to the ferroelectric behaviour, and in
where a large part of the work was carried sized materials. Some scientists have general give novel functions to materials.”
out – as well as Tuhin Chakrabortty, PhD proposed compelling theories about why
student at CeNSE. and how these materials switch, but they Directly imaging the dynamics of light
have not been experimentally proven so elements using electron microscopy has not
Using state-of-the-art atomic resolution far. been attempted until recently. Among the
microscopy, the researchers show many challenges was fabricating electron
experimentally for the first time how Nukala and colleagues used an advanced microscope-compatible capacitors made
ferroelectricity emerges in materials called electron microscopy technique that had of hafnia. The task of designing these was
hafnia-based oxides. It arises from the recently been developed and earlier used successfully led by Nukala, who believes that
displacement and reversible movement of by a research team at the University of the protocols they have developed could help
negatively charged oxygen atoms when an Groningen to visualise a hydrogen atom, researchers perform similar experiments on
electric field is applied. Such materials are the lightest chemical element. In the new other materials.
useful for low-power memory applications, study, they imaged thin films of hafnium-
according to Nukala, who is one of the zirconium oxide sandwiched between Another challenge was accurate statistical
corresponding authors of the study published two electrodes. They were also able to analysis of the microscopy data. Nukala’s
in Science. “Hafnia-based ferroelectric track the movement of atoms, including student Chakrabortty, who has been working
memory devices are already in production, oxygen, in real time when an electric on statistical inference in biological systems,
even though the mechanism behind their field was applied. used image processing tools to establish that
behaviour was not known,” he says. the reversible oxygen migration is indeed
The researchers found that charged statistically significant.
Like magnets, ferroelectric materials show oxygen atoms move from one electrode
spontaneous polarisation – the separation to another with the hafnia layer acting The insights offered by the study open up
of positive and negative charges – which can as a conduit. When the electric field new avenues for designing oxygen-conducting
be reversed or switched using an electric was reversed, the direction of migration ferroelectric materials that could be used for
field. However, they are generally unsuitable was also reversed. It was this migration miniature memory and logic devices.
for miniaturisation because they lose their that contributed significantly to the
ferroelectric properties when the crystal is material’s ferroelectricity, they found.
made smaller than a particular size. When the conduit size was reduced - Ranjini Raghunath
KERNEL | ISSUE 6, 2021 | Pg 3
Tracking the Thatcher effect in deep networks trained on objects vs trained on faces (Image: Adapted from Jacob et al./Nature Communications)
DO DEEP NETWORKS “SEE”
AS WELL AS HUMANS?
UNDERSTANDING HOW MACHINE LEARNING MODELS MEASURE UP TO HUMANS CAN HELP SCIENTISTS MAKE
THEM MORE ROBUST
A new study from the Centre for attempted to understand which visual Similarly, when presented with an image
Neuroscience (CNS) at IISc explores how tasks can be performed by these networks of a face, humans first look at the face as
well deep neural networks compare to naturally by virtue of their architecture, a whole, and then focus on finer details
the human brain when it comes to visual and which require further training. like the eyes, nose, mouth and so on,
perception. explains Georgin Jacob, first author and
The team tested 13 different perceptual PhD student at CNS. “Surprisingly, neural
Deep neural networks are machine effects on the deep networks. An example networks showed a local advantage,” he
learning systems inspired by the network is the Thatcher effect, a phenomenon where says. This means that unlike the brain, the
of brain cells or neurons in the human humans find it easier to recognise local networks focus on the finer details of an
brain, which can be trained to perform feature changes in an upright image, but image first. Therefore, even though these
specific tasks. These networks have more challenging when the image is flipped neural networks and the human brain
played a pivotal role in helping scientists upside-down. Deep networks trained to carry out the same object recognition
understand how our brains perceive recognise upright faces showed a Thatcher tasks, the steps followed by the two are
the things that we see. Although deep effect when compared with networks very different.
networks have evolved significantly over trained to recognise objects. Another
the past decade, they are still nowhere visual property of the human brain, called “Lots of studies have been showing
close to performing as well as the human mirror confusion, was tested on these similarities between deep networks and
brain in perceiving visual cues. In a recent networks. To humans, mirror reflections brains, but no one has really looked at
study, SP Arun, Associate Professor at along the vertical axis appear more similar systematic differences,” says Arun, who is
CNS, and his team have compared various than those along the horizontal axis. The the senior author of the study. Identifying
qualitative properties of these deep researchers found that deep networks these differences can push us closer to
networks with those of the human brain. also show stronger mirror confusion for making these networks more brain-like.
vertically-reflected images as compared to
Deep networks, although a good model horizontally-reflected images. Such analyses can also help researchers
for understanding how the human brain build more robust neural networks that
visualises objects, work differently from Another feature peculiar to the human not only perform better but are also
the latter. While complex computation brain is that it first focuses on coarser immune to adversarial attacks that aim to
is trivial for them, certain tasks that details. This is known as the global derail them.
are relatively easy for humans can be advantage effect. For example, when we
difficult for these networks to complete. are shown an image of a tree, our brains - Sangeetha Devi Kumar
In the current study, published in Nature would first see the tree as a whole before
Communications, Arun and his team noticing the details of the leaves in it.
KERNEL | ISSUE 6, 2021 | Pg 4
Image: Santanu Mukherjee
NOVEL SYNTHESIS OF DIFFICULT-TO-MAKE
INDOLE DERIVATIVES
Indoles and compounds derived from two distinct forms whose 3D structures Fischer indolization. Computational studies
indoles are found in many natural are non-superimposable mirror images of through Density Functional Theory (DFT)
substances. These compounds have each other, called enantiomers. Often, only calculations, led by Garima Jindal, helped
extensive applications in pharmaceuticals one enantiomer is biologically relevant, in shedding light on the mechanism of this
and agrochemicals. but it is extremely difficult to selectively reaction.
synthesise just one of them individually in
A research team from the Department the lab – most existing methods produce a Using this method, the team synthesised
of Organic Chemistry led by Santanu mixture of both. a range of enantioenriched cyclopenta[b]
Mukherjee has now developed an efficient indolones which are challenging to make.
method to synthesise cyclopenta[b] The IISc team has overcome this hurdle Their method, therefore, increases the
indolones – an important class of indole using a catalyst which is also chiral in scope of making indole derivatives.
derivatives. These molecules are ‘chiral’ nature, and by carrying out a modified
in nature which means they can exist in version of a century-old reaction called - Sritama Bose
Image: Kaling Danggen, Shivangi Mishra and Varsha Singh
USING ROUNDWORMS TO
STUDY SKIN DISEASES
Collagen proteins in our skin play an skin responds to environmental toxins. were found to be more susceptible to
important role in preventing environmental Throughout its lifetime, it produces 177 herbicides and deworming drugs like
toxins from entering our body. Defects in different types of collagens and some ivermectin. The loss of four out of the six
this collagen barrier can cause several of these are important for maintaining crucial proteins led to physical changes
diseases like Gaucher’s disease, atopic its skin shape and structure. But little is in the skin structure, making it look
dermatitis and psoriasis. known about the function of a majority of discontinuous and wrinkled under a high
these 177 proteins. resolution electron microscope. They
A new study from Varsha Singh’s lab at the also found that when skin permeability
Department of Molecular Reproduction, In the study, the researchers found that at was compromised, it led to increased
Development and Genetics proposes least six of these proteins are responsible accumulation of toxins, greater tissue
that Caenorhabditis elegans, a soil- for determining what can pass through damage and accelerated death.
dwelling roundworm, may be used as a the skin. Genetically modified worms
model organism to understand how our lacking even one of these proteins - Debayan Dasgupta
KERNEL | ISSUE 6, 2021 | Pg 5
Photo: Roshan R Rao
TRADITIONAL DWELLINGS ARE BETTER
SUITED FOR CHANGING CLIMATE
Many people have gradually shifted loggers, they recorded temperatures inside climatic zone, traditional dwellings were
from traditional houses – which rely the houses every 30 minutes for almost a warmer indoors, making them more
on locally available building materials year, and built a mathematical model to suitable for residence. In the warm-humid
and knowledge – to modern dwellings, predict how indoor temperatures would and temperate climatic zones, modern
even in rural areas. Khadeeja Henna, be in the future. The team then simulated houses had relatively higher indoor
Aysha Saifudeen and Monto Mani from three future global warming scenarios with temperatures. This would make them more
the Centre for Sustainable Technologies different levels of greenhouse gas emissions, dependent on artificial air conditioning,
recently studied which of the two were and estimated how the houses fared. fuelling global warming further. The study
more resilient to climate change. suggests that traditional dwellings have
In all three climates, traditional houses design solutions that can help adapt
They evaluated houses in three different – such as ones with timber walls or slate better to climate change.
Indian villages that had temperate, roofing – were less affected by climate
warm-humid and cold climates. Using data change than modern houses. In the cold - Joel P Joseph
Image: Ishika Pramanick and Nayanika Sengupta
NOVEL INSIGHTS INTO THE STRUCTURE OF
SARS-CoV-2 SPIKE PROTEIN
The SARS-CoV-2 S proteins, which appear infects the host cells – remains poorly conformation at physiological pH 7.4, but
as crown-like spikes on the viral surface, understood. their proportion decreases when the pH
mediate the entry of the virus into the is slightly higher or lower.
host cell. They are also the site where Researchers in the Molecular
neutralising antibodies produced by the Biophysics Unit led by Somnath Dutta They also detected many intermediate
host cells bind to the virus to inactivate have now successfully visualised the conformations between fully open and
it. Understanding the protein’s detailed different conformations or forms of the closed, and showed that distinct states
structure is therefore important. S protein at physiological pH (7.4) and of both conformations have different
near physiological pH (6.5 and 8.0) binding affinities towards neutralising
Most previous studies on the S protein using a technique called single-particle antibodies.
structure have been carried out either at pH cryo-electron microscopy.
8.0 or pH 4.0 to pH 5.0. But the structure - Sritama Bose
of the S protein at physiologically relevant The team observed that around
conditions – at which the virus actually 68% of the S proteins exist in open
KERNEL | ISSUE 6, 2021 | Pg 6
Photo courtesy: Sanjiv Sambandan
WHEN ELECTRONICS
FLEXES ITS MUSCLES
SANJIV SAMBANDAN’S RESEARCH PROMISES NOVEL ELECTRONIC DEVICES THAT CAN BEND AND FOLD
During Sanjiv Sambandan’s undergraduate on such flexible substrates hold enormous questions led to the genesis of
days as an electrical engineering student promise: they have the potential to enable Openwater.in, a startup that
at IIT Kharagpur, he was introduced to the futuristic devices with rollable displays, specialises in treating wastewater.
world of flexible electronics, especially e-paper, smart wearable sensors and It employs a technology that builds
the work being done at the University transparent Radio-frequency ID tags. on electrocoagulation, a well-known
of Waterloo in Canada, where he would technique used to treat waste suspended
eventually pursue his PhD. His research was In the quest for flexible electronic systems, in water. Electrocoagulation involves
on flexible electronics on plastic substrates, Sanjiv’s lab tries to take advantage of the changing the particle surface charge,
an idea that he thought was “very cool.” impact of geometrical, mechanical and causing the suspended particles
Following his doctoral studies, he worked at electrical stresses on thin film devices, a to agglomerate, which can then be
Xerox PARC in California, where he explored phenomenon that other engineers usually separated. Using insights from this
a diverse range of research, paving the way regard as a problem in flexible systems. research, Sanjiv’s lab came up with
for a genuinely interdisciplinary academic He and his team have been exploring a cost-effective process of enhancing
career. how this handicap can be turned into an electrocoagulation, ensuring that
advantage. In this process, they have agglomerates are formed more
In 2010, he decided to return to academia invented devices and circuit techniques for aggressively and at lower power.
and applied for a faculty position at IISc. flexible electronics, including the adaptive
“At some point, the academic bug begins dielectric thin film transistor, corrugated Since 2017, Openwater.in, supported by
to bite,” recalls Sanjiv, who got the job and thin-film transistor and self-compensating grants from the Government of India’s
returned to his hometown, Bangalore, as circuits. Department of Biotechnology, has been
an Assistant Professor in the Department of able to build systems that treat up to
Instrumentation and Applied Physics (IAP). Another aspect of Sanjiv’s lab involves 10,000 L of wastewater per day. The
interdisciplinary research that studies the startup is currently involved in several
Sanjiv’s IAP lab attempts to solve behaviour of dispersions in electric fields, government projects and has plans to
fundamental problems of flexible biodegradable electronic materials and so set up several groundwater, industrial
electronics and semiconductor physics. on. This research has led to innovations discharge and grey water treatment
His group uses this research to build thin- such as self-healing circuits and self- plants across the country, including one
film transistor-based integrated systems. assembled sensors and actuators. Such at Mavallipura that can generate 25,000
Flexible electronics does not use wafers interdisciplinary research has also led to L of potable water per day. Although
like silicon or gallium arsenide, commonly technology translation. A shining testament Openwater.in is working exclusively
employed in conventional electronics. to this is the lab spinoff that aims to treat with the government, several apartment
Instead, the main focus here is to make wastewater. complexes have also sought the help of
thin-film transistors by depositing a layer the startup to treat their grey water.
of patterned semiconductor on flexible One of Sanjiv’s students, Aswathi Nair,
materials like textiles and plastics that can explored how dispersions behave in an Sanjiv’s research on flexible electronics
bend and fold. Electronic circuits printed electric field for her MTech thesis. These has also received widespread attention,
KERNEL | ISSUE 6, 2021 | Pg 7
Waste water treatment unit developed by Openwater (Photo courtesy: Openwater.in)
particularly his recent work on developing Another fascinating foray for Sanjiv’s lab has Sanjiv, now Associate Professor, is delighted
“self-healing” circuits. In flexible been in, believe it or not, edible electronics. with what he has achieved as a researcher
materials, stretching causes circuits Edible electronics uses benign, and often at IISc. “I have been an electrical engineer
to break, thus reducing the lifespan of food-grade, materials like sugar, proteins doing interdisciplinary work. All of it has
wearable electronics. To help address the and silver thin films to form circuits. His been encouraged and taken positively here.”
problem, he mooted the idea of electronic team aims to use this technology to make
circuits which repair themselves. The edible pills that can perform diagnostic tests A decade-long career as a researcher at IISc
self-healing mechanism kicks in to repair to yield a ‘yes’ or a ‘no’ answer, without has also seen Sanjiv mentor several students
open circuit faults by using dispersions of supervision from a doctor. Sanjiv believes from all over India. Based on his experience
oil and metallic microparticles smeared that this will serve a niche area of diagnosis as a mentor, he has a message for those just
over connections between various between traditional invasive tests that need starting their research careers: “Explore a
circuit elements. If a fault occurs in the a doctor’s precious time and wearable fitness lot. Do not restrict yourselves. Be creative
connections in the circuit, the electric field gadgets that are often inaccurate in their like a child.” He also urges students to go
polarises the metallic particles and forces diagnosis. However, the challenge for his back to the basics of their subject of interest.
them to bridge the gap. The particles are lab in the coming years will be to integrate “Focus strongly on the fundamentals.”
fixed in place by sintering once the flow of sub-systems and make these pills safe for
current is restored. humans to ingest. - Debayan Dasgupta
Self healing circuits (Image courtesy: Sanjiv Sambandan)
Office of Communications (OoC) EDITORIAL TEAM DESIGN
Indian Institute of Science (IISc) Deepika S TheFool.in
Karthik Ramaswamy
Bengaluru - 560012
Narmada Khare
kernel.ooc@iisc.ac.in Ranjini Raghunath (Coordinator)
https://kernel.iisc.ac.in/ Samira Agnihotri
KERNEL | ISSUE 6, 2021 | Pg 8
You can also read
