Dr. Anoushka Joglekar, NY Genome Center, presents seminar at UGA Bioinformatics

00:46:33
https://www.youtube.com/watch?v=eR3SbNc8agw

概要

TLDREl Dr. Jer ha presentat la seva recerca sobre les isoformes d'ARNm i l'alternativa del splicing en cèl·lules individuals. En la seva xerrada, ha destacat com aquestes isoformes poden influir en la salut i la malaltia, evidenciant exemples com el gen cd33 en relació amb l'Alzheimer. A més, ha parlat de la metodologia desenvolupada pel seu laboratori per estudiar aquestes isoformes a través d'una tècnica anomenada 'scissor-seq'. També va mostrar dades sobre com aquestes isoformes poden variar entre tipus de cèl·lules i durant el desenvolupament del cervell. Jer ha subratllat la importància d'estudiar l'expressió específica d'isoformes tant en la investigació bàsica com en el desenvolupament de fàrmacs, destacant que fins a un 75% de gens poden expressar diferents isoformes depenent del tipus de cèl·lula.

収穫

  • 👨‍🔬 Dr. Jer és un expert en genoma 3D i isoformes d'ARN.
  • 🧬 L'alternativa del splicing afecta la salut i malalties com l'Alzheimer.
  • 🔬 Desenvolupament de la tècnica 'scissor-seq' per estudiar isoformes.
  • 🧠 Les isoformes varien entre tipus cel·lulars i durant el desenvolupament del cervell.
  • 📊 75% dels gens poden expressar diferents isoformes en funció del tipus de cèl·lula.
  • 🔍 Comprendre quin isoforma està expressada és crític per al desenvolupament de medicaments.
  • 🧪 L'estudi se centra en teixit cerebral, explorant etapes del desenvolupament.
  • ⚠️ Cal investigar més sobre el comportament de les isoformes al llarg del temps.
  • 🌍 Importància de traduir recerques de models de ratolí a humans.
  • 📚 Conferència valuosa per a professionals en biologia computacional i genòmica.

タイムライン

  • 00:00:00 - 00:05:00

    Introducció de la Dr. Jer, investigadora en el Centre del Genoma de Nova York, que desenvolupa algoritmes per explorar l'estructura 3D del genoma. Presentació del seu treball en esplicació alternativa amb resolució de cèl·lula única.

  • 00:05:00 - 00:10:00

    La Dr. Jer parla sobre les isoformes d'ARN i la seva importància en la salut i la malaltia. Explica el procés d'esplicació alternativa i la seva influència en la complexitat del proteoma.

  • 00:10:00 - 00:15:00

    Discussió sobre l'esplicació alternativa com a específica de teixit, fins i tot dins d'un sol teixit. Descripció de la metodologia per estudiar les isoformes en cèl·lules individuals.

  • 00:15:00 - 00:20:00

    Explicació del mètode Scissor-seq per superar les limitacions del seqüenciament basat en gotes, permetent l'identificació d'isoformes transcriptòmiques completes en cèl·lules individuals.

  • 00:20:00 - 00:25:00

    Presentació dels estudis sobre teixit cerebral per examinar la variabilitat de les isoformes entre regions cerebrals i tipus de cèl·lules. Exemples de gens específics estudiats.

  • 00:25:00 - 00:30:00

    Troballes sobre la variabilitat de les isoformes entre cèl·lules i regions cerebrals, especialment en exons micro. Discussió d'exemples de diferències observades.

  • 00:30:00 - 00:35:00

    Investigació de la persistència de les observacions d'isoformes en diverses regions cerebrals al llarg del temps. Mètodes per categoritzar la variabilitat de les isoformes.

  • 00:35:00 - 00:40:00

    Estudi detallat de gens amb gran variabilitat d'isoformes en diversos tipus cel·lulars i etapes del desenvolupament. Exploració de patrons específics de canvis en les isoformes.

  • 00:40:00 - 00:46:33

    Conclusions sobre l'estudi de les exons extremadament variables i com les diferències en isoformes estan relacionades amb tipus cel·lulars específics i etapes de desenvolupament.

もっと見る

マインドマップ

Mind Map

よくある質問

  • Qui és el Dr. Jer?

    El Dr. Jer és un investigador postdoctoral al New York Genome Center.

  • Quin era l'objectiu de la xerrada del Dr. Jer?

    Discutir la importància de l'alternativa del splicing en cèl·lules individuals i el seu impacte en salut i malalties.

  • En quin camp es centra la recerca del Dr. Jer?

    En el desenvolupament d'algoritmes per explorar l'estructura tridimensional del genoma i estudiar les isoformes d'ARNm.

  • Què és l'alternativa del splicing?

    És un procés intermediari que ocorre entre la transcripció d'ADN a ARN i la traducció d'ARN a proteïna, que pot influir significativament en funcions cel·lulars i salut.

  • Per què és important estudiar les isoformes d'ARNm segons el Dr. Jer?

    Perquè les diferents isoformes poden tenir impactes significatius en salut i malalties, alterant la funció i estructura de les proteïnes.

  • Quina tècnica ha desenvolupat el laboratori del Dr. Jer per estudiar les isoformes?

    Han desenvolupat una tècnica anomenada 'scissor-seq' per obtenir lligams d'ARN complet en cèl·lules individuals.

  • Quin és un exemple d'un gen estudiat pel Dr. Jer?

    Un dels gens estudiats és binan, implicat en malalties com l'Alzheimer.

  • El Dr. Jer considera que les isoformes són un punt focal en el desenvolupament de drogues?

    Sí, argumenta que comprendre quina isoforma està expressada és crucial en el desenvolupament de medicaments.

ビデオをもっと見る

AIを活用したYouTubeの無料動画要約に即アクセス！
字幕
en
オートスクロール:
  • 00:00:09
    very awesome good morning everyone it's
  • 00:00:12
    my pleasure to introduce Dr Jer as
  • 00:00:14
    today's seminar speaker um Dr J lier is
  • 00:00:18
    currently a postdoc researcher at New
  • 00:00:21
    York genome Center where she's
  • 00:00:23
    developing algorithms to explore the 3D
  • 00:00:26
    structure of genome across different
  • 00:00:28
    cell types she got her PhD in
  • 00:00:31
    computation biology and medicine from
  • 00:00:33
    whale Coral medicine at Corell
  • 00:00:36
    University working in Tingler lab her
  • 00:00:39
    doctoral research focused on creating
  • 00:00:41
    algorithms and Technologies for studying
  • 00:00:43
    alternative splicing in the brain at
  • 00:00:46
    single cell and single nucleus
  • 00:00:48
    resolution I've had the opportunity to
  • 00:00:51
    work with a few of the tools that she
  • 00:00:53
    has developed and they are really great
  • 00:00:56
    um I especially I'm especially grateful
  • 00:00:58
    to her for agreeing without visitation
  • 00:01:00
    when I invited her to present at today's
  • 00:01:03
    seminar so please join me in welcoming
  • 00:01:06
    Dr
  • 00:01:09
    juger um very nice to meet you in person
  • 00:01:14
    for the first time n yeah and uh thank
  • 00:01:17
    you for inviting me I'm very excited to
  • 00:01:19
    give this talk and um before I begin I
  • 00:01:22
    just want to say that everyone is
  • 00:01:24
    welcome to interrupt me at any point uh
  • 00:01:28
    I'd like to keep this as informal as
  • 00:01:30
    possible uh but of course we can save um
  • 00:01:34
    questions for the end as well so I'm
  • 00:01:36
    going to aim for this to be 40 to 45
  • 00:01:38
    minutes but if I run over just stop me
  • 00:01:40
    because sometimes that happens okay so I
  • 00:01:44
    actually have to figure out how to do
  • 00:01:46
    this um given that I've never shared a
  • 00:01:51
    keynote presentation before let's see
  • 00:01:55
    okay you see this right yeah we do see
  • 00:01:58
    see your screen okay and do you see like
  • 00:02:01
    um the full screen mode of the uh
  • 00:02:05
    presentation okay awesome
  • 00:02:07
    thanks okay so before we jump into the
  • 00:02:12
    actual science I wanted to First
  • 00:02:15
    discuss what um isoforms are why we
  • 00:02:19
    should even study them in single cells
  • 00:02:22
    what's the point of it all
  • 00:02:25
    so excuse me okay so we all know the
  • 00:02:30
    Dogma where we go from a DNA to an RNA
  • 00:02:33
    molecule um and the functional unit most
  • 00:02:36
    people will say is protein but for us
  • 00:02:39
    transcriptomics people it's the Gene and
  • 00:02:42
    uh this process of going from DNA to RNA
  • 00:02:45
    is called transcription and then from
  • 00:02:47
    RNA you get protein um in a process
  • 00:02:50
    called
  • 00:02:51
    translation and often times what is
  • 00:02:54
    ignored is there's an intermediate step
  • 00:02:58
    between um transcript description and
  • 00:03:00
    translation and that step is called
  • 00:03:02
    alternative splicing so the way that
  • 00:03:06
    works is that you have a very complex
  • 00:03:10
    um like a complex col spical complex and
  • 00:03:15
    it's a collection of proteins it's got
  • 00:03:16
    about a 100 or so proteins and what it
  • 00:03:19
    does is that it spices out um introns uh
  • 00:03:23
    or junk DNA from your or junk RNA from
  • 00:03:27
    your RNA molecule and along with the
  • 00:03:29
    anons it also spices out certain exons
  • 00:03:32
    so you get like this combinatorial
  • 00:03:35
    expansion of um isoforms these are
  • 00:03:39
    called RNA isoforms which then get
  • 00:03:42
    translated into protein isoforms so as I
  • 00:03:45
    said it like combinatorially expands the
  • 00:03:49
    proteome and uh one of my favorite like
  • 00:03:52
    illustrations of this is that the worm
  • 00:03:56
    mice and humans we all approximately
  • 00:03:58
    have the same gene which is why we use
  • 00:04:00
    worm and mice as model organisms right
  • 00:04:04
    and I would argue that as we go down
  • 00:04:07
    this trajectory we get increasingly
  • 00:04:10
    complex and one source of this
  • 00:04:12
    complexity is the fact that each um Gene
  • 00:04:16
    and a worm is only represented as one
  • 00:04:19
    isopor versus in humans there can be up
  • 00:04:22
    to 10 sometimes hundreds of isoforms per
  • 00:04:25
    Gene and if this hasn't convinced you um
  • 00:04:28
    isoforms aren't just like tweaking the
  • 00:04:31
    proteins a little bit they actually do
  • 00:04:33
    make uh quite big impact in health and
  • 00:04:36
    disease
  • 00:04:38
    so here is a uh Gene called cd33 and
  • 00:04:43
    it's a microgo so like an immune Gene
  • 00:04:46
    and the M like capital M isopor has two
  • 00:04:51
    um alternative exons which are included
  • 00:04:54
    and the small m is um not the canonical
  • 00:04:58
    version and that this Exon and that
  • 00:05:02
    actually um is neuroprotective in terms
  • 00:05:05
    of Alzheimer's disease and the first one
  • 00:05:07
    is neurotoxic so depending on what
  • 00:05:09
    isopor you have it literally makes a
  • 00:05:12
    difference between health and disease in
  • 00:05:13
    some cases life and death um there's
  • 00:05:16
    different types of alternative splicing
  • 00:05:18
    and the one that I just showed and what
  • 00:05:20
    is the most commonly seen in mammals is
  • 00:05:23
    Exxon skipping so you either have or
  • 00:05:25
    don't have excuse me this red isopor and
  • 00:05:28
    then you can have a bunch of other types
  • 00:05:31
    uh depending on whether you're skipping
  • 00:05:33
    two at a time or you have a shorter Gene
  • 00:05:36
    because of the first or the last
  • 00:05:38
    isoforms being included or
  • 00:05:41
    excluded so splicing can be tissue
  • 00:05:44
    specific uh meaning that you can have
  • 00:05:46
    the same gene have two different
  • 00:05:47
    isoforms and say the brain versus the
  • 00:05:49
    gut and as you know genes don't occur in
  • 00:05:53
    like a isolated form they have Gene Gene
  • 00:05:56
    interactions protein protein
  • 00:05:57
    interactions so it can have a lot of
  • 00:05:59
    Downstream effects based off of where
  • 00:06:01
    your isopor is expressed and then um
  • 00:06:05
    it's not just tissue specific even
  • 00:06:07
    within tissues it can have differences
  • 00:06:10
    so that's kind of what I was studying
  • 00:06:12
    and in my case I was studying the brain
  • 00:06:14
    but there's other tissues such as the
  • 00:06:16
    liver that make use of isoforms quite
  • 00:06:19
    frequently for regulation so there can
  • 00:06:21
    be different regions of the brain
  • 00:06:23
    expressing different isoforms and to
  • 00:06:25
    make matters more complicated you can
  • 00:06:27
    have different cell types um when I
  • 00:06:30
    started my PhD this was new so this
  • 00:06:32
    slide is now not quite out of date but
  • 00:06:34
    everyone knows um how single cell is
  • 00:06:37
    studied and usually speaking there's
  • 00:06:40
    some degree of droplet based
  • 00:06:42
    microfluidics um and that's like the way
  • 00:06:45
    to get High throughput um single cell
  • 00:06:47
    RNA signal but of course there's other
  • 00:06:49
    methods and um you would generally
  • 00:06:52
    represent that in a tne uh one second
  • 00:06:56
    I'm just going to skip a couple slides
  • 00:06:57
    because we don't need these
  • 00:07:00
    uh one thing I do want to say is that um
  • 00:07:03
    droplet based RNA seek while capturing
  • 00:07:06
    single cell information doesn't capture
  • 00:07:09
    the full information and the reason for
  • 00:07:10
    that is that it is three or a five Prime
  • 00:07:14
    quantification method for the most part
  • 00:07:16
    smart seek or other versions aren't just
  • 00:07:19
    three prime Quant but um 10x for example
  • 00:07:23
    pars these um have three or five Prime
  • 00:07:27
    bias what that means from an isopor
  • 00:07:29
    perspective if you look at the full um
  • 00:07:33
    formulation what you'll find is that you
  • 00:07:35
    know you could maybe be sequencing this
  • 00:07:38
    end of your read and you could maybe
  • 00:07:40
    tell that this is a protein C versus
  • 00:07:43
    protein A or B being expressed but for
  • 00:07:46
    the most part you wouldn't really be
  • 00:07:47
    able to tell the difference between
  • 00:07:49
    proteins A and B if you got you know 10x
  • 00:07:52
    3 Prime sequencing for example so to
  • 00:07:55
    circumvent this our lab came up the
  • 00:07:58
    tilgner lab came up with
  • 00:07:59
    um method called scissor seek and it's
  • 00:08:02
    very simple it's really not rocket
  • 00:08:05
    science the way it works is you follow
  • 00:08:07
    your um short rate sequencing Paradigm
  • 00:08:10
    the way you normally would and what we
  • 00:08:13
    do is we reserve half of our um cdna and
  • 00:08:18
    we do this pref fragmentation so what
  • 00:08:21
    that means is that we can send half of
  • 00:08:24
    that cdna the same like like a different
  • 00:08:26
    aload of the same sample for long rate
  • 00:08:29
    sequencing
  • 00:08:30
    and because the long raid will contain
  • 00:08:32
    the entire isopor but also contain the
  • 00:08:35
    barcodes um we can figure out which
  • 00:08:38
    barcode is coming from which single cell
  • 00:08:41
    and then you can basically identify um
  • 00:08:44
    and Trace back fulllength transcripts to
  • 00:08:47
    their cell cell and then we usually work
  • 00:08:50
    on a cell type level because of
  • 00:08:52
    throughput um so you can get cell type
  • 00:08:55
    specific
  • 00:08:56
    isoforms so this happened in like my
  • 00:08:59
    first first ISS of my PhD and then once
  • 00:09:02
    we had this method up and running we
  • 00:09:04
    wanted to use it um on brain tissue
  • 00:09:09
    so what we wanted to do was find out um
  • 00:09:13
    a couple things firstly we looked at a
  • 00:09:16
    very young so postnatal day seven which
  • 00:09:18
    is think of it like as a baby mouse um
  • 00:09:22
    and we wanted to find out if there are
  • 00:09:25
    two different brain regions say the
  • 00:09:26
    hippocampus and the prefrontal cortex
  • 00:09:29
    which which are both important for
  • 00:09:30
    memory cognition all of the high level
  • 00:09:33
    processes that we have in mammals um if
  • 00:09:37
    you look at these two different brain
  • 00:09:38
    regions firstly for the same cell type
  • 00:09:41
    let's say asites found between
  • 00:09:43
    hippocampus and PFC do we see
  • 00:09:46
    differences in isopor I said that there
  • 00:09:48
    was Regional variability but to what
  • 00:09:50
    extent do you find it and then the
  • 00:09:52
    second thing was within a single brain
  • 00:09:54
    region what is the extent of isopor
  • 00:09:57
    variability between different cell types
  • 00:09:59
    is it that you know like 90% of the
  • 00:10:02
    genes are expressing the same isopor in
  • 00:10:05
    different cell types or is it like um
  • 00:10:08
    much much lower and isoforms play a
  • 00:10:11
    pretty big regulatory role so I'm not
  • 00:10:14
    going to give too much detail about this
  • 00:10:16
    but the punch line is that there is a
  • 00:10:19
    lot of variability uh both between and
  • 00:10:22
    within regions for the between regions
  • 00:10:25
    we found about 400 genes that are pretty
  • 00:10:28
    highly implicated
  • 00:10:30
    um in Regional variability and we found
  • 00:10:32
    like changes on the level of what we
  • 00:10:35
    call microexons so like 15 to 30 base
  • 00:10:37
    pairs that can change the protein
  • 00:10:40
    structure um and therefore probably
  • 00:10:42
    change the function um I'll just focus
  • 00:10:45
    on an example in the hippocampus so we
  • 00:10:48
    looked at not just within cell types
  • 00:10:51
    like asites versus neurons but we looked
  • 00:10:55
    at um within neuronal types so here I'm
  • 00:10:59
    going to show you a gene called fgf13
  • 00:11:02
    it's a growth Gene it peaks in
  • 00:11:04
    expression right around this p7 PA time
  • 00:11:08
    point and what I'm showing you here is a
  • 00:11:10
    heat map on the like rows we have two
  • 00:11:15
    isoforms the S is the short and the VY
  • 00:11:18
    is the long isopor and notice that the
  • 00:11:21
    only difference between them is where
  • 00:11:23
    this one starts versus this one starts
  • 00:11:25
    so this is happening on the five Prime
  • 00:11:27
    end not something you would be able to
  • 00:11:29
    see using just 10x sequencing and it's
  • 00:11:32
    colored by the degree to which the
  • 00:11:35
    isopor is expressed so we call it Pi Pi
  • 00:11:38
    per inclusion of an isopor and what
  • 00:11:41
    you'll notice is that the excitatory
  • 00:11:44
    neurons which are this middle section
  • 00:11:46
    here
  • 00:11:47
    exclusively um Express the short isopor
  • 00:11:51
    versus the other neurons and
  • 00:11:53
    particularly I want to draw your
  • 00:11:54
    attention to V uh inhibitory neurons
  • 00:11:57
    they modulate the expr expression of
  • 00:11:59
    both of them to varying degrees and when
  • 00:12:03
    we did some overexpression studies what
  • 00:12:05
    we found was that the S isopor not only
  • 00:12:09
    is it expressed differently in different
  • 00:12:11
    cell types but where it's expressed
  • 00:12:13
    where it's localized within the cell in
  • 00:12:15
    this case it's the nucleis in this case
  • 00:12:17
    it's the entire cytoplasm um that also
  • 00:12:20
    makes a difference so you definitely
  • 00:12:23
    know that there is some degree of
  • 00:12:25
    functional uh variation depending on
  • 00:12:28
    which isopor is being expressed and this
  • 00:12:30
    is not just for like big cell types but
  • 00:12:33
    at a pretty granular granular level of
  • 00:12:36
    cell subtypes um this is unsolicited
  • 00:12:40
    advice but this was this um whole
  • 00:12:43
    example came about because I had a
  • 00:12:45
    friend called Susan well she wasn't
  • 00:12:48
    really a friend I didn't know her but
  • 00:12:49
    she saw me give a talk uh about this in
  • 00:12:52
    like an internal forum and she was like
  • 00:12:55
    oh would you mind looking at this Gene
  • 00:12:58
    because I think they're could be
  • 00:12:59
    something here um so I encourage you
  • 00:13:03
    during your PhD talk to your friends
  • 00:13:05
    give as many talks as you can um that
  • 00:13:08
    way something cool can come out of it um
  • 00:13:10
    for both you and your uh
  • 00:13:13
    collaborator so this is um within the
  • 00:13:17
    brain regions and then another thing we
  • 00:13:19
    did to sort of validate the between
  • 00:13:22
    brain region thing was spatial
  • 00:13:24
    sequencing was just coming out so 10x
  • 00:13:26
    had this like beta version of their
  • 00:13:30
    um what's Zoom platform and we got
  • 00:13:33
    access to it so what we wanted to do was
  • 00:13:35
    see if um or the signal that we found of
  • 00:13:39
    regional variability in isoforms could
  • 00:13:41
    we see that using slide seek or whatever
  • 00:13:43
    visium Plus long rate sequencing and
  • 00:13:47
    indeed we find that to be true so snap
  • 00:13:50
    25 is another developmental Gene it
  • 00:13:53
    Peaks again in expression actually I
  • 00:13:56
    don't think it peaks in expression what
  • 00:13:58
    happens is that there's a known switch
  • 00:14:00
    uh from isopor A to B that happens at
  • 00:14:04
    this p7 p8 um time point in postnatal
  • 00:14:08
    development and um we find that this
  • 00:14:11
    switch is very very visible just at a
  • 00:14:15
    snapshot in time so the a isopor again
  • 00:14:18
    I'm showing you a very similar plot to
  • 00:14:20
    before except it it's not a heat map
  • 00:14:21
    anymore um each dot here represents a
  • 00:14:24
    spot on the visium um slide and yellow
  • 00:14:28
    means that it's expressed highly black
  • 00:14:31
    means that it's expressed lowly and so
  • 00:14:33
    the young isopor or the young Exxon is
  • 00:14:37
    expressed in the front of the brain and
  • 00:14:39
    the old Exxon or like older um isopor is
  • 00:14:43
    expressed in the back of the brain and
  • 00:14:45
    this is kind of a known thing that the
  • 00:14:47
    brain sort of develops in a back to
  • 00:14:51
    front manner um with the cerebellum
  • 00:14:54
    becoming more active as you can see like
  • 00:14:56
    you know kids don't have that much
  • 00:14:58
    prefrontal cortex or cognitive
  • 00:15:00
    capabilities that's because it's kind of
  • 00:15:02
    slower um and so you can really see that
  • 00:15:05
    dynamism in action just even at a single
  • 00:15:08
    snapshot Okay so we've established at
  • 00:15:11
    this time Point um and you can read more
  • 00:15:14
    about it in this paper um that isoforms
  • 00:15:18
    change on a brain region level isoforms
  • 00:15:21
    change on a cell type level and they
  • 00:15:23
    change on a cell subtype level but now
  • 00:15:25
    there's more questions that remain um so
  • 00:15:29
    so first thing do these observations
  • 00:15:31
    that we made at the postnatal time point
  • 00:15:33
    do they persist a along different brain
  • 00:15:37
    regions b along different time points
  • 00:15:40
    and the main question and I don't think
  • 00:15:43
    I've gotten even close to answering it
  • 00:15:46
    but in humans what happens because we
  • 00:15:48
    use mice as model organisms a lot but if
  • 00:15:52
    you can't use what you've learned from
  • 00:15:55
    mice and like extrapolate that to human
  • 00:15:58
    what is the point
  • 00:15:59
    um or can we can we finesse um the genes
  • 00:16:02
    in which we can actually model things
  • 00:16:04
    versus we cannot so this last part I
  • 00:16:08
    won't talk about too much given time
  • 00:16:10
    constraints but if we do get to it um
  • 00:16:13
    I'll talk about how we optimize the
  • 00:16:16
    scissor seek protocol to be used in
  • 00:16:17
    Frozen tissue but if I don't get to it
  • 00:16:20
    um you can read about it in this paper
  • 00:16:23
    okay
  • 00:16:25
    so what we did to ask about this
  • 00:16:28
    persistence over regions as well as time
  • 00:16:31
    is we uh designed a study wherein we
  • 00:16:34
    would kind of sequence everything we
  • 00:16:36
    would sequence space which is different
  • 00:16:38
    brain regions at an adult time point and
  • 00:16:41
    two brain regions the visual cortex and
  • 00:16:44
    the
  • 00:16:45
    hippocampus um overdevelopment and the
  • 00:16:48
    reason we chose this was because we were
  • 00:16:50
    part of the a Consortium and they really
  • 00:16:53
    wanted to focus on particular brain
  • 00:16:55
    regions excuse me but I think this was a
  • 00:16:58
    very good pan
  • 00:16:59
    because uh these two brain regions
  • 00:17:02
    capture a lot of like cognitive and
  • 00:17:04
    memory and all this development versus
  • 00:17:07
    these are more
  • 00:17:09
    um like they they take care of a lot of
  • 00:17:12
    involuntary processes and Regulation and
  • 00:17:15
    things like that okay so we got these 11
  • 00:17:20
    time points and we sequence single cell
  • 00:17:22
    samples from there um and there were two
  • 00:17:25
    replicates for each so we ended up with
  • 00:17:26
    about 200,000 cells which again now
  • 00:17:29
    doesn't seem like a lot but when we
  • 00:17:31
    started definitely did and um this is
  • 00:17:34
    just a you map showing what um the cell
  • 00:17:38
    types were and um what I want to point
  • 00:17:41
    out is that this is like the entire data
  • 00:17:44
    set right but there's so much to explore
  • 00:17:47
    how do we go about exploring it so one
  • 00:17:50
    thing we did was we first broke it down
  • 00:17:52
    by uh time point and brain region so you
  • 00:17:55
    have this brain regions um showing
  • 00:17:58
    different um colocalization and whatever
  • 00:18:01
    you map things which you don't want to
  • 00:18:03
    read too much into but you can see the
  • 00:18:05
    different brain region specific say
  • 00:18:08
    neurons for example and you can do the
  • 00:18:10
    same for age so if we want to study how
  • 00:18:14
    not like how the gene expression changes
  • 00:18:16
    but how isoform expression changes um
  • 00:18:19
    there's different axes of variability of
  • 00:18:22
    this so you can have the region like I
  • 00:18:25
    said the age or you can have cell
  • 00:18:27
    subtype and so so we wanted to go about
  • 00:18:29
    this in first like a very broad Strokes
  • 00:18:33
    kind of way to see what are the broad
  • 00:18:36
    changes brought about by development and
  • 00:18:39
    brain region um uh changes so how we
  • 00:18:44
    measured this was I'm going to go
  • 00:18:46
    through an illustrative example and then
  • 00:18:48
    go into actual examples so let's
  • 00:18:52
    consider Gene called XYZ or whatever and
  • 00:18:56
    there's five isoforms for the Gene and
  • 00:19:00
    this is a given cell type so it's it's
  • 00:19:02
    not
  • 00:19:03
    overall um what we can do is we can
  • 00:19:06
    measure that pi value that I said the
  • 00:19:08
    percent isopor and we can measure that
  • 00:19:11
    along all three axes so the first one is
  • 00:19:13
    region and then you can do the same for
  • 00:19:15
    cell um subtype and time and what we do
  • 00:19:20
    is we to measure the variability we just
  • 00:19:23
    look at the max of the Delta Pi in most
  • 00:19:26
    cases you're going to have a pretty un
  • 00:19:28
    uniform expression or that's what we
  • 00:19:30
    assumed um but in some cases you might
  • 00:19:33
    have one or two outliers that are very
  • 00:19:36
    different in how they express a certain
  • 00:19:38
    isopor and this was informed by our
  • 00:19:40
    previous study so you can find this Max
  • 00:19:43
    Delta Pi for the region and the other
  • 00:19:45
    two axes and um then you can represent
  • 00:19:50
    that as a vector so um 30 20 35 in this
  • 00:19:54
    case is the change in the Delta pi and
  • 00:19:58
    you can do the same for not just this
  • 00:20:00
    isopor but all isoforms and so you can
  • 00:20:02
    have a matrix instead of a vector and
  • 00:20:05
    what we did was we not R normalized it
  • 00:20:08
    and it has a couple benefits and a
  • 00:20:10
    couple um disadvantages so what you can
  • 00:20:14
    do is you can represent this in a Turner
  • 00:20:17
    plot or a Simplex whatever you want to
  • 00:20:19
    call it so it's kind of like a scatter
  • 00:20:21
    plot except it has three axes instead of
  • 00:20:23
    Two And depending on which triangle it
  • 00:20:28
    ends up in you know that that particular
  • 00:20:31
    isopor is either subtype or region or
  • 00:20:35
    age specific um the center triangle is
  • 00:20:38
    this like kind of complic it's not
  • 00:20:39
    complicated but it's like this weird
  • 00:20:41
    place where it can either be all low or
  • 00:20:44
    all high because we have normalized it
  • 00:20:47
    and there's no real way of telling them
  • 00:20:48
    apart just visually but what we were
  • 00:20:51
    interested in which axis is like the
  • 00:20:54
    most um important for a given cell type
  • 00:20:59
    so what we find is that excitatory
  • 00:21:01
    neurons here each uh dot is an isopor
  • 00:21:05
    and this is true for like I don't know
  • 00:21:07
    12 or 15,000 genes um they all sort of
  • 00:21:12
    like colocalize to the bottom triangle
  • 00:21:15
    so this is a subtype triangle and um and
  • 00:21:19
    it's not just the subtype right so what
  • 00:21:21
    we look at is is a gene uh if a gene has
  • 00:21:25
    multiple isoforms are there are they all
  • 00:21:29
    in the subtype triangle and so that's
  • 00:21:30
    like a gene expression change or are
  • 00:21:33
    they in multiple triangles meaning that
  • 00:21:35
    this is true isopor regulation and so
  • 00:21:38
    that's something I've represented in
  • 00:21:39
    this like Network diagram and what we
  • 00:21:42
    find is that for most genes the subtype
  • 00:21:46
    variability is usually accompanied by
  • 00:21:50
    something else um one of the other axes
  • 00:21:53
    but subtype variability is the most
  • 00:21:56
    predominant and so we find this very
  • 00:21:58
    exory neurons and when you switch when
  • 00:22:00
    you go into other cell types you find
  • 00:22:03
    that uh the pattern looks different so
  • 00:22:05
    in this case it's mostly on the bottom
  • 00:22:07
    right here it's like the bottom and here
  • 00:22:09
    it's the top and left so really
  • 00:22:12
    different cell types have different ways
  • 00:22:15
    in which their isoforms change and this
  • 00:22:18
    would be what I would call the broad
  • 00:22:19
    strokes and then what you can do is you
  • 00:22:22
    can sort of zoom in on individual um
  • 00:22:26
    genes to see like you know are there
  • 00:22:29
    genes that Express variability across
  • 00:22:31
    different cell types and across
  • 00:22:33
    different axes so I isolated a few I
  • 00:22:36
    forget how many I'm sorry but here is a
  • 00:22:39
    gene called ruy 3 and I call this a
  • 00:22:42
    hypervariable gene and the reason for
  • 00:22:44
    that will become apparent in a minute so
  • 00:22:48
    here are six seven seven isoforms for
  • 00:22:52
    this Gene and what I'm going to show you
  • 00:22:55
    is um for a given cell type this
  • 00:22:57
    triangle plot that I showed you
  • 00:23:00
    previously um and below that triangle
  • 00:23:02
    plot I'm going to also show you what the
  • 00:23:04
    isopor percentage so Pi uh that I showed
  • 00:23:08
    you
  • 00:23:09
    before is for different ages and here
  • 00:23:12
    you'll see that for the most part um
  • 00:23:15
    it's kind of distributed but for this
  • 00:23:17
    purple um isopor the p28 time point is
  • 00:23:21
    kind of like
  • 00:23:23
    special uh so this is true for
  • 00:23:25
    progenitors and then if you look at some
  • 00:23:27
    other cell type you'll see that the heat
  • 00:23:29
    map starts looking different so I point
  • 00:23:31
    out a couple things for example
  • 00:23:33
    excitatory neurons at all ages seem to
  • 00:23:36
    prefer this second isopor versus
  • 00:23:39
    astrocytes at all ages seem to prefer
  • 00:23:42
    this pink isopor and again if you like
  • 00:23:45
    start staring more and more at this
  • 00:23:47
    you'll notice that oh it's not just the
  • 00:23:49
    fact that it's an isopor change it's
  • 00:23:50
    like how is that affecting the protein
  • 00:23:52
    how are the ends being folded in the
  • 00:23:55
    protein structure and you can like
  • 00:23:56
    really go down a rabbit hole
  • 00:23:59
    depending on what Gene you're looking at
  • 00:24:01
    another thing I want to point out is
  • 00:24:02
    that in the immune cells you go from one
  • 00:24:06
    isopor being used to a different isopor
  • 00:24:09
    being used to again some diffus sort of
  • 00:24:11
    structure so if there's anything you can
  • 00:24:15
    take away from this is that these
  • 00:24:17
    isoforms can get really really
  • 00:24:18
    complicated really really fast and
  • 00:24:21
    that's true for not just age but regions
  • 00:24:24
    as well as cell types and um don't
  • 00:24:27
    ignore your forms guys is is all I'm
  • 00:24:30
    saying um and you can find this
  • 00:24:32
    information like you can explore these
  • 00:24:34
    genes in more detail on this isopor
  • 00:24:37
    atlas.com um but the point is that we
  • 00:24:41
    found something like 75% of genes
  • 00:24:44
    Express um different isoforms across at
  • 00:24:47
    least one axis which is kind of crazy so
  • 00:24:51
    not only do different cell types have
  • 00:24:53
    different genes being expressed but 75%
  • 00:24:56
    of genes have a different isopor
  • 00:24:58
    depending on which cell type you're
  • 00:24:59
    looking at okay so this is the broad
  • 00:25:02
    Strokes version now we want to zoom in
  • 00:25:04
    so what we did was we looked at
  • 00:25:06
    something called extremely variable
  • 00:25:07
    exons and the reason for that is as you
  • 00:25:10
    can see these different isoforms have so
  • 00:25:12
    many moving Parts how do you know what's
  • 00:25:15
    what's changing so we focused on
  • 00:25:17
    individual exons and uh the way we did
  • 00:25:20
    this was this is just for illustrative
  • 00:25:22
    purposes but let's say you have two uh
  • 00:25:26
    cell types like one cell types cell type
  • 00:25:28
    present across two different brain
  • 00:25:31
    regions in adulthood what you can do is
  • 00:25:34
    you can measure the change in Exxon
  • 00:25:36
    usage across these two brain regions and
  • 00:25:40
    um I would call that brain region
  • 00:25:42
    specific difference and if that change
  • 00:25:45
    is more than 75% I think that was our
  • 00:25:48
    arbitrary cut off then we report that as
  • 00:25:51
    something being extremely variable for
  • 00:25:54
    adult brain region specific differences
  • 00:25:56
    you can do the same with the
  • 00:25:59
    um like a map for different cell types
  • 00:26:02
    as well as within a developmental time
  • 00:26:04
    point so we're calling that
  • 00:26:05
    developmental cell type uh differences
  • 00:26:08
    and finally you can do it over time um
  • 00:26:11
    like how is the isopor landscape
  • 00:26:14
    changing within an olgod dendrite
  • 00:26:17
    population over time so we isolated
  • 00:26:21
    these extremely variable exons and that
  • 00:26:23
    came out to be about 600 something exons
  • 00:26:27
    Manning fewer genes and if you notice
  • 00:26:31
    straight off the bat there's this one
  • 00:26:33
    giant cluster in the middle and what
  • 00:26:36
    that cluster is is it's both adult and
  • 00:26:39
    developmental cell type spe specific
  • 00:26:41
    differences now this is not a surprising
  • 00:26:44
    thing because you do know like we do
  • 00:26:47
    know that big cell types like neurons
  • 00:26:50
    versus glea are going to have pretty
  • 00:26:53
    large differences regardless of which
  • 00:26:55
    time point or which brain region you
  • 00:26:57
    look at and um that is exactly what we
  • 00:27:01
    found uh in this network diagram I'm
  • 00:27:03
    showing where the differences came from
  • 00:27:05
    so the on the right are neurons and the
  • 00:27:07
    left are Gia and as you can see there's
  • 00:27:10
    nothing connecting the neurons and
  • 00:27:12
    nothing connecting the Gia meaning that
  • 00:27:14
    this is Neuron versus Gal changes that
  • 00:27:17
    we're seeing the most unsurprising but
  • 00:27:20
    the cool thing about it is like we went
  • 00:27:22
    and looked at not just particular genes
  • 00:27:25
    but the protein domains um that these in
  • 00:27:28
    code and it's funny that there's like
  • 00:27:31
    these uh fibronectin type domains which
  • 00:27:33
    are highly repetitive regions and even
  • 00:27:36
    within them you can see that two exons
  • 00:27:39
    are being used by neurons and two exons
  • 00:27:40
    are being used by Gia um if I understand
  • 00:27:44
    that correctly yep and um so like just
  • 00:27:49
    it shouldn't make a difference that
  • 00:27:52
    these highly repetitive structures have
  • 00:27:54
    some method to their Madness but it does
  • 00:27:57
    so that is all the more reason for
  • 00:27:59
    people to look at not just which protein
  • 00:28:02
    like the canonical form but which
  • 00:28:04
    isoform is being used especially in like
  • 00:28:07
    drug uh development for example okay so
  • 00:28:11
    the next cluster we looked at was this
  • 00:28:13
    one over here in the middle of these two
  • 00:28:16
    and what we see is that there is adult
  • 00:28:19
    cell type specific difference as well as
  • 00:28:21
    adult brain region specific difference
  • 00:28:24
    so what that means is what we're finding
  • 00:28:26
    is for a given Gene in this case it's
  • 00:28:29
    called
  • 00:28:30
    tex9 um we find that a given isopor is
  • 00:28:35
    or Exon is different for a brain region
  • 00:28:39
    and for a particular cell type so in
  • 00:28:42
    this case we're seeing that an ISO um
  • 00:28:44
    Exxon is going from 90% inclusion to 10%
  • 00:28:48
    11% inclusion depending on which brain
  • 00:28:51
    region this is coming from I should also
  • 00:28:53
    mention that this is a uh plot kind of
  • 00:28:57
    that we developed in house on the bottom
  • 00:29:00
    is each line represents um the Gen code
  • 00:29:03
    annotation and on the top each line
  • 00:29:05
    represents a single read coming from
  • 00:29:08
    whatever this uh region is so and I
  • 00:29:12
    think n has used that as well and
  • 00:29:14
    hopefully without too many issues but um
  • 00:29:17
    this is kind of how we show differences
  • 00:29:19
    in cell types or brain regions or
  • 00:29:21
    whatever our thing of interest is okay
  • 00:29:25
    so this is what we find that whenever
  • 00:29:28
    have a brain region specific difference
  • 00:29:30
    it is usually occurring in single cell
  • 00:29:32
    types as opposed to all cell types and
  • 00:29:34
    this kind of agrees with what we'd found
  • 00:29:37
    um in the postnatal mouse brain as well
  • 00:29:40
    the next um thing we look at is these
  • 00:29:42
    two um are lit up so they're
  • 00:29:45
    developmental um changes but they are
  • 00:29:49
    age specific and in a given cell type so
  • 00:29:51
    again it's a same as the brain region
  • 00:29:53
    but this is happening over development
  • 00:29:56
    and here you'll see that for the
  • 00:29:58
    hippocampus you're going from like a 70%
  • 00:30:01
    to a 40% to 100% inclusion of a given
  • 00:30:05
    Exxon but that's not really true for say
  • 00:30:08
    a given for the prefrontal cortex or
  • 00:30:10
    sorry visual
  • 00:30:11
    cortex um another thing that I'm just
  • 00:30:15
    going to touch upon and you're going to
  • 00:30:16
    be like where do human come from um was
  • 00:30:19
    we sequenced human uh tissue kind of on
  • 00:30:23
    the side but we basically found out that
  • 00:30:26
    whenever we see cell type specific spe
  • 00:30:28
    it it is conserved in human but when you
  • 00:30:32
    find cell type specificity in human um
  • 00:30:35
    it is not always found in Mouse so that
  • 00:30:38
    kind of ties back into what I was saying
  • 00:30:40
    before like you can model things from
  • 00:30:42
    Mouse to human but the reverse Direction
  • 00:30:44
    I Would caution against I'm not going to
  • 00:30:47
    go over this plot for time purposes um
  • 00:30:49
    so let's move on and then the last plot
  • 00:30:52
    I want to last cluster I want to talk
  • 00:30:54
    about is this one this to me was the
  • 00:30:57
    most surpris in what we're seeing is
  • 00:30:59
    that there's changes that are occurring
  • 00:31:02
    in development that are not cell type
  • 00:31:04
    specific and they kind of disappear in
  • 00:31:07
    adulthood and I was very intrigued by
  • 00:31:09
    what this could be so we found a gene uh
  • 00:31:12
    that we've looked at before and which is
  • 00:31:14
    what I'm going to focus on now it's
  • 00:31:15
    called binan it's a neuronal like um
  • 00:31:19
    like brain specific Gene it's been
  • 00:31:21
    implicated in Alzheimer's disease a lot
  • 00:31:24
    and we've studied it pretty heavily in
  • 00:31:26
    the past so we we looked at P1 which is
  • 00:31:30
    like really tiny baby mice and um this
  • 00:31:34
    is again a scissor Wiz plot um wherein
  • 00:31:37
    you have your annotation on the bottom
  • 00:31:40
    and different cell types and different
  • 00:31:41
    colors the orange represents um
  • 00:31:45
    alternative exons so what you'll find is
  • 00:31:49
    that um these neuronal types the ones in
  • 00:31:52
    blue happen to include the orange a lot
  • 00:31:56
    and then the cell um types below which
  • 00:31:59
    are the G types do include the orange
  • 00:32:02
    but not that much so in our head we were
  • 00:32:06
    like this is a gal sorry this is a gal
  • 00:32:09
    isopor and this is a neuronal isopor and
  • 00:32:12
    that's just how it is for bin one we
  • 00:32:15
    look at our like new data and what we
  • 00:32:17
    find is this so at um
  • 00:32:20
    p14 depending on whether you're looking
  • 00:32:22
    at um hippocampus or visual cortex you
  • 00:32:27
    will see that um this neuronal isopor is
  • 00:32:31
    present where you have some of the green
  • 00:32:33
    which is orange here um even both cell
  • 00:32:36
    types and then at p56 but this is true
  • 00:32:40
    for all time points in all good endr
  • 00:32:42
    sites which are glea you have the G
  • 00:32:44
    isopor cool as the cells or mice mature
  • 00:32:49
    you find truly that this is there like
  • 00:32:52
    you see the neuronal isopor um but then
  • 00:32:56
    when we look at P2 21
  • 00:32:59
    hippocampus it starts to kind of
  • 00:33:02
    disappear right like the green is not as
  • 00:33:05
    as strongly expressed as it was as it is
  • 00:33:08
    over here then we go to p28 and the
  • 00:33:11
    green almost entirely disappears so what
  • 00:33:15
    that is meaning and we're looking at
  • 00:33:16
    this across multiple replicates and
  • 00:33:19
    multiple genes and things like that um
  • 00:33:22
    the green is disappearing so the neurons
  • 00:33:25
    have taken on almost like a GLE identity
  • 00:33:28
    of their isopor and I don't have a
  • 00:33:31
    question like answer for why but I do
  • 00:33:33
    know that this is happening and what's
  • 00:33:36
    even crazier is that it kind of bounces
  • 00:33:38
    back so it's not just monotonically
  • 00:33:42
    increasing or decreasing overdevelopment
  • 00:33:45
    it is literally going up and down so
  • 00:33:47
    then I wanted to look at um what is how
  • 00:33:51
    it's going up and down um so I started
  • 00:33:56
    like trying to do this in in a bit more
  • 00:33:59
    um systematic manner so what I looked at
  • 00:34:04
    was instead of isopor variability um
  • 00:34:07
    across the different axes I looked at it
  • 00:34:10
    just over time and what I'm looking at
  • 00:34:14
    is cell type variability so if all cell
  • 00:34:17
    types uh have a given variability here
  • 00:34:20
    for p14 and a different variability at
  • 00:34:23
    p21 then I measure the Delta between
  • 00:34:26
    them
  • 00:34:28
    um I I can do the same for p21 to
  • 00:34:32
    p28 and then I can do the same for p28
  • 00:34:35
    to p56 so whenever there's an
  • 00:34:38
    outlier um say in this case this red
  • 00:34:41
    point asite is showing differences at
  • 00:34:43
    p28 that would be captured because that
  • 00:34:47
    um variability is pretty high and then
  • 00:34:50
    it's kind of Disappearing at a later
  • 00:34:51
    time point so for most genes you would
  • 00:34:55
    expect that things are going pretty
  • 00:34:57
    stable uh but you can have genes like
  • 00:35:01
    this where it's kind of not that
  • 00:35:03
    different at this um transition from p14
  • 00:35:07
    to p21 but from p21 to 28 it's kind of
  • 00:35:11
    going up and then it's going down at
  • 00:35:14
    p56 so you can see this pattern um as
  • 00:35:18
    you can imagine there are 27 possible
  • 00:35:20
    patterns because there's three at every
  • 00:35:23
    what's it called um
  • 00:35:26
    interval and um so we tried to basically
  • 00:35:31
    figure out how many genes follow which
  • 00:35:34
    pattern so as I mentioned most of the
  • 00:35:37
    genes or most of the exons are
  • 00:35:39
    invariable they're going to be pretty
  • 00:35:42
    stable um across the different cell
  • 00:35:44
    types at um each interval but then
  • 00:35:48
    there's about a thousand genes uh Exon
  • 00:35:51
    that show a very like different types of
  • 00:35:54
    patterns so first I'm going to show you
  • 00:35:56
    the invariant one because it's easy to
  • 00:35:58
    show and then I'll show you the others
  • 00:36:01
    so over here I'm just showing line plots
  • 00:36:04
    with the green being like the average
  • 00:36:07
    and um this number on the y- axis is
  • 00:36:11
    showing the variability so as you can
  • 00:36:13
    see for the most part it's like not
  • 00:36:15
    changing that much and the average is
  • 00:36:17
    about the same and you can visualize the
  • 00:36:19
    actual um Exon expression value and
  • 00:36:23
    again you can see that it's fairly green
  • 00:36:24
    or fairly red either all low or all high
  • 00:36:28
    but then you can look at some of these
  • 00:36:30
    other patterns that are showing up
  • 00:36:32
    especially the the Thousand xon are show
  • 00:36:35
    so that I showed and what you'll see is
  • 00:36:37
    what I showed you with the bin one
  • 00:36:39
    example where it's not just going up or
  • 00:36:42
    not just coming down it's kind of going
  • 00:36:45
    up and down in Crazy different ways and
  • 00:36:48
    that's not just like this variability
  • 00:36:50
    that I'm showing but you can see that
  • 00:36:52
    like in individual exons if you
  • 00:36:55
    visualize the uh VAR ility PSI and the
  • 00:37:00
    other thing I want to point out is that
  • 00:37:02
    if you look at these patterns the uh two
  • 00:37:05
    time points that come out the most but
  • 00:37:07
    the one time point that comes out the
  • 00:37:09
    most is p28 so it's always the
  • 00:37:12
    transition between p21 to 28 that is
  • 00:37:14
    either going like um up or down or being
  • 00:37:19
    crazy um so and the reason for that and
  • 00:37:22
    this is my hypothesis it's speculation
  • 00:37:24
    that's not been tested is that the p21
  • 00:37:28
    to 28 time point is when is called the
  • 00:37:31
    critical developmental time point in
  • 00:37:33
    mice it is after mice have opened their
  • 00:37:36
    eyes and they have like started to
  • 00:37:39
    actually see the world uh so in human
  • 00:37:42
    terms it's days not like also days U but
  • 00:37:46
    it's kind of crazy that there's things
  • 00:37:48
    happening on a developmental front in
  • 00:37:51
    the brain that you're seeing in action
  • 00:37:54
    um for individual
  • 00:37:56
    exons okay so with that I'll uh conclude
  • 00:38:00
    this part actually I'll conclude my talk
  • 00:38:02
    because I'm basically out of time um so
  • 00:38:06
    as you as I showed you we looked at
  • 00:38:08
    extremely variable exons and what we
  • 00:38:11
    found was that the cell type specific
  • 00:38:14
    differences may it be developmental or
  • 00:38:16
    may it be adult um those are the most
  • 00:38:20
    prevalent what we find is that the adult
  • 00:38:23
    brain region specific differences are
  • 00:38:25
    usually accompanied by cell type
  • 00:38:28
    specific differences so there is some
  • 00:38:31
    like critical modulation happening um of
  • 00:38:36
    um critical modulation happening of
  • 00:38:39
    function on a cell type specific level
  • 00:38:42
    uh even though it's across different
  • 00:38:43
    brain regions the next thing we found
  • 00:38:46
    was that it was conserved in human and
  • 00:38:49
    um lastly we found that uh these
  • 00:38:52
    developmental differences that we found
  • 00:38:54
    that are critical are transient they
  • 00:38:57
    they don't like they don't come and go
  • 00:39:00
    in a monotonic fashion they come and
  • 00:39:03
    then they disappear especially around
  • 00:39:06
    this critical uh time point so this work
  • 00:39:11
    was recently published in April uh and
  • 00:39:14
    you're welcome to go read it or ask me
  • 00:39:17
    questions whatever um I also wanted to
  • 00:39:20
    talk to you about developing single
  • 00:39:22
    nucleus isopor sequencing but
  • 00:39:24
    unfortunately I don't have that much
  • 00:39:26
    time so just going to go straight to my
  • 00:39:28
    acknowledgement slide um I'd like to
  • 00:39:32
    thank the tner lab which is where I did
  • 00:39:34
    most of this work um and when and Bay
  • 00:39:38
    were the ones who did most of the sample
  • 00:39:40
    processing for this project the brain
  • 00:39:42
    initiative was the funding body for it
  • 00:39:46
    and um of course lots of
  • 00:39:49
    collaborators and um I actually switched
  • 00:39:52
    from the New York genome Center but I
  • 00:39:54
    worked with um the ulinski and the gors
  • 00:39:57
    Labs while I was there and so I want to
  • 00:39:59
    thank them as well and um I'm happy to
  • 00:40:02
    take any questions you guys may
  • 00:40:14
    have thank you great talk Dr jar um I
  • 00:40:17
    have a few questions but I feel like um
  • 00:40:20
    we can open the floor for others to ask
  • 00:40:22
    first yeah sure
  • 00:40:33
    okay U While others are thinking I'll
  • 00:40:35
    I'll I'll get started uh anishka again
  • 00:40:37
    fantastic talk I really enjoyed it um so
  • 00:40:40
    um I mean it's this is fascinating you
  • 00:40:42
    know during the developmental stage and
  • 00:40:44
    know how how the different isop forms
  • 00:40:46
    are changing I was curious if with with
  • 00:40:49
    any of these isop forms do you see sort
  • 00:40:50
    of correlated changes in in the other
  • 00:40:53
    isop forms because ultimately we won't
  • 00:40:55
    understand this at sort of the network
  • 00:40:56
    context right so so have have you have
  • 00:40:59
    you found any correlated changes with
  • 00:41:02
    any of these that's a great question and
  • 00:41:05
    I would love to explore that especially
  • 00:41:07
    in like you know because like Gene
  • 00:41:09
    regulatory networks are going to have
  • 00:41:12
    these corresponding isopor regulatory
  • 00:41:14
    networks and I haven't saided them uh
  • 00:41:17
    personally but that's a that's a um
  • 00:41:20
    something on the radar because we wanted
  • 00:41:22
    to actually look at and I don't know if
  • 00:41:25
    they're doing this still but we wanted
  • 00:41:27
    to look at how RNA binding protein
  • 00:41:30
    changes affect these so that would be
  • 00:41:33
    the best way to study the coordination
  • 00:41:35
    we did look because of reviewer comments
  • 00:41:37
    so it was a very brief look um at
  • 00:41:42
    something like that where like if you
  • 00:41:44
    have changes like a knockdown in an RBP
  • 00:41:47
    how that affects multiple exons or um
  • 00:41:50
    genes at once but we didn't do a not a
  • 00:41:54
    satisfactory analysis so they don't have
  • 00:41:57
    a good answer okay well thank
  • 00:42:17
    you
  • 00:42:20
    Jessica hiy yeah thank you um really
  • 00:42:22
    nice talk um I've done Isa seek
  • 00:42:25
    interestingly in single cellular
  • 00:42:27
    organisms where you don't have different
  • 00:42:28
    tissues and we are seeing different
  • 00:42:30
    isoforms one of the challenges we
  • 00:42:32
    encountered and it's just a
  • 00:42:34
    methodological uh question here detail
  • 00:42:37
    um did you rely on the reverse
  • 00:42:39
    transcription to get to the five Prime
  • 00:42:41
    end or do you do cap selection you know
  • 00:42:45
    yeah no we didn't do cap selection This
  • 00:42:47
    was um we just did RT and we definitely
  • 00:42:51
    see so like I don't know if you uh I'll
  • 00:42:54
    try and pull up a oops that was a
  • 00:42:56
    different one um I'll try and pull up a
  • 00:42:59
    plot so for example actually just in the
  • 00:43:02
    bin one case if you look on the left
  • 00:43:05
    this was generated with pack bio and uh
  • 00:43:09
    this was generated with on and as you
  • 00:43:13
    can see like already there is like um
  • 00:43:17
    differences in how full length the
  • 00:43:20
    molecule is so this is on a
  • 00:43:22
    technological front but also with the RT
  • 00:43:25
    we see a similar thing where if you do
  • 00:43:27
    human samples sometimes you'll have
  • 00:43:30
    random stuff happening and it doesn't go
  • 00:43:32
    to the five Prime end of the molecule
  • 00:43:35
    and so that's just kind of
  • 00:43:39
    um like
  • 00:43:40
    a yeah you just kind of have to deal
  • 00:43:43
    with it artifactual thing um is what we
  • 00:43:47
    ended up on but that being said I don't
  • 00:43:49
    know if you followed work from the crg
  • 00:43:53
    but they did do some cap trap new method
  • 00:43:57
    that does capture the five Prime tail
  • 00:43:59
    and apparently that is much better at
  • 00:44:02
    getting the entire molecule um yeah yeah
  • 00:44:06
    yeah I mean that's kind of the
  • 00:44:07
    conclusion that I had come to because I
  • 00:44:09
    was just seeing a lot of variability and
  • 00:44:10
    I didn't know if they degradation
  • 00:44:12
    project you know products or just our te
  • 00:44:14
    coming off you know like you just you
  • 00:44:15
    don't know what's going on and so you
  • 00:44:17
    know having that five Prime cap
  • 00:44:19
    selection seems to be important all
  • 00:44:21
    right thank you
  • 00:44:30
    I have a quick question so when you say
  • 00:44:32
    alternative splicing is there a
  • 00:44:34
    threshold that you that the software
  • 00:44:36
    uses that saying if this percentage of
  • 00:44:40
    um the cells cell population has this
  • 00:44:42
    Exon that means it's an alternative SP
  • 00:44:45
    alterntive yeah uh no that's a good
  • 00:44:47
    question and there's two parts to it so
  • 00:44:49
    we generally work on a cell type level
  • 00:44:52
    and the reason for that is that these
  • 00:44:54
    long read methods still don't have the
  • 00:44:57
    same level of throughput as say alumina
  • 00:44:59
    so you're getting like a tenth or so of
  • 00:45:02
    the throughput so we don't work on a
  • 00:45:04
    cell level um so we don't do that filter
  • 00:45:09
    but once we get to a point where we're
  • 00:45:10
    getting millions and millions of reads
  • 00:45:13
    that is definitely something to do what
  • 00:45:15
    we do do is um we ask for at least a 10%
  • 00:45:21
    change between cell types um to classify
  • 00:45:25
    it as an alternative spicing
  • 00:45:28
    event um and that can change so we've
  • 00:45:31
    we've played around with this like we
  • 00:45:33
    don't just use the Delta we also use a P
  • 00:45:37
    value because like you get more
  • 00:45:38
    confidence in your results if you kind
  • 00:45:40
    of use two Avenues as opposed to just
  • 00:45:42
    one um and so sometimes like if we want
  • 00:45:45
    to be hyper confident that's why I use
  • 00:45:47
    the 75% cut off in the alternative sorry
  • 00:45:50
    extremely variable exons because we
  • 00:45:52
    wanted to be sure that this is something
  • 00:45:54
    that we like a robust signal that we're
  • 00:45:57
    but in most cases it'll be 10 sometimes
  • 00:46:09
    20% um if there are no more questions
  • 00:46:12
    then thank you Dr Jer it was a great
  • 00:46:15
    talk um and thank you everyone for
  • 00:46:18
    joining I'll connect with you offline
  • 00:46:20
    for more of my questions by that sounds
  • 00:46:23
    good yeah okay you awesome thank you bye
  • 00:46:26
    bye hi thank you
タグ
  • Dr. Jer
  • genoma 3D
  • isoformes ARN
  • alternativa del splicing
  • salut i malaltia
  • scissor-seq
  • cèl·lules individuals
  • desenvolupament cerebral
  • teixit cerebral
  • Alzheimer