Видео

The guy deserves a lot of respect for a clear and simple explanation of complex things. Great job!

Would love to see a ChUG podcast on "You're doing it wrong..but so was I". Where flow folks discuss things like only using 10% bleach and not using a detergent. The common misconceptions we had because "it was the way you I was taught", and they were such basic things you never bothered to question them.

Algorithms do imact the comensation/unmixing. I unmixed same data using diferent algorithms and the output was different.

I great discussion about flow cytometry companies

Only thing better than the content on Chug is David's beard.

Great content! I enjoyed every minute 👍👍

Another great chug podcast!

I tested the restrictor on my LSR II after watching this and found it was leaking! Likely the cause of some time delay issues we were having!

An excellent discussion, really enjoyed it. Maybe you can discuss the topic of sample preparation in future. Things that can go wrong, steps that people do not pay attentions etc.

If Imaging is a marketing add on (and a BD exclusive add-on at that), I think the critical question is does it beat other exclusive add-ons from other companies? Personally I cant shake the idea that a flow cytometer without acoustic focusing is going to really suffer (particularly with imaging flow cytometers). Image sorting on a BD Discover OR non-sorting imaging on a future Cytek-Imagestream OR a Bigfoot with the Cytpix?

Just came to know about Deepcell, any thoughts? Looks like image based sorter, label free.

Hello! Great video! Just a comment but the (fitc pe) matrix should be vertical to be correctly representing the matrix multiplication

Super happy that I found ChUG videos! After going through endless hours of flow material on the internet, where in an hour at best there is one sentence that is actual news to me, your videos are full of things that I always wanted to know, or at least things that are confirming what I thought.

This was fun! Finally I found something I used to know better than you: I actually knew how to pronounce Chattopadhyay , on account that a fellow postdoc had the same name. Now all my advantage is gone :)

Suppose I'm that "one and half person" individual that was in Louisville. Good insight here.

I am a big fan and follower of Pratip. An excellent researcher and one of the guiding lights in flow Cytometry. Whenever I get stuck thinking about how to plan my next move in my career and science I try looking for what Pratip is doing- you can say I follow his steps. Really appreciate this podcast. I am a staff scientist in NIH at Flow Cytometry core in NIH -Bethesda campus.

An excellent source to clear confusion...

Here’s how I think about dimensionality reduction vs clustering: Dimensionality reduction is the easiest way to view all 40 parameters at the same time. Let’s say I have a dataset with two groups: healthy controls and patients with a disease. I want an overview to visualize the differences between my groups. With 40 parameters it's going to be tedious to make tons of biaxial plots to investigate all parameters and attempt to get a global overview. So I perform dimensionality reduction on all of the data, and then separate it to create two plots - one plot shows all the healthy data and one plot shows the patient data. From this I can instantly tell if there are major differences - maybe my patients are missing 3 islands compared to the healthy controls. The dimensionality reduction (without clustering) is a tool to display data and in my experience it’s most useful when comparing global differences between two or more groups of samples. However, most people want to know what cell types/populations are different between the groups of samples. That’s where clustering comes in. Once cells have been clustered, we can then overlay that onto the plot of dimensionally reduced data. We could do some calculations with just the clustering information to determine what’s different between healthy vs patient, but if I’ve already got the dimensionally reduced data it’s much easier to visualize the differences. Maybe we put the clustering data on top of the dimensionally reduced data plot and we instantly find that clusters 2, 7, and 10 are the ones missing in the patient samples. Then I can go examine those specific clusters further. Because clustering is like dividing a pie into multiple slices, we don’t necessarily have to have two or more groups of samples for it to be useful. But combining the clustering and dimensionality reduction can be really helpful. One last note about dimensionality reduction (without clustering) is that it’s much more impactful in published figures when population/island differences are major, like if they are present or absent. It’s like when you zoom out really far on an image and you can’t see any minor details. If you instead find that the frequency of the Tregs in patients is 10% compared to 5% in healthy controls, it may be statistically significant data but a tSNE plot probably isn’t going to be the best way to communicate that specific discovery. Hope that helps! - Laura

One point was mentioned that if the fluorophore gets attached to the hinge region of the Ab it will destroy the Ab over time. How can control that? Is it possible to predict?

Really great conversation...

We will not bring appropriate controls but Compensation MUST work is loose equivalent to car wash! @22 min

Great discussion guys, can you bring Shapiro? That will be fun

$UNSTAINEDCENTERS is the MFI (or other central tendency) of an unstained sample, which can be used for autofluorescence subtraction. The example from the publication is: $UNSTAINEDCENTERS♦3,FL1-A,FL2-A,FL3-A,123.45,234.56,65.4321♦ So, there are three $PnN with central tendencies of 123.45, 234.56, and 65.4321. Not sure if/why you need to list the $PnN if they are just ordered by the N.

Easy to understand matrix algebra presentation. Thanks!

Well done Dave

Very nice presentation! Congratulations.