As a master student I have to thank you for your videos. Without them, getting into the analysis of RNAseq-data would be a major chore. Thank you and stay healthy!
Hey man, your videos about different RNA-seq topics and problems are more than awesome, I (and probably others) really appreciate this. Keep up the good work with this and your other StatQuest videos!
+Joshua Starmer Just one question about this video (I'm new to this): When you say 'read-counts', what exactly are you referring to? The numbers of reads of length k obtained from a gene X, during an RNA-seq experiment? And that number is different for each individual (ignoring the technical problems) because of different gene expression, even if the individuals are genetically identical? And also, is the combined length of all reads from a single gene X equal to the length of the whole gene X? I.e. the whole expressed gene is sequenced, or are some parts left out? Sorry if I'm bothering, couldn't find a nice explanation and was hoping you could provide a little insight whenever you have time.
Please add explanatory video about stranded rna seq and their protocols in illumina because I could never understand the rna seq clustering and amplification step for that. Also about paired end reads using examples of sequences because its hard to understand the reverse complement and complement reads trhings.
Hiii. I would be glad to read a paper about this. Yesterday I was searching and I didn't find any paper talking about biological vs technical replicates in RNA-seq. If I have to make a decision, I would do it based on a paper
Hi Josh! Love your song🥲. If I passage the cells grow in one dish into 3 dishes, Let them grow or give them the same type of treatment. Harvest each RNA from one dish as one sample. Send that 3 RNA samples to seq. Is these biological sample or technical samples? Thank you !
Thanks. you mean that the biological variation cancels out the technical variation?? and that is why we should have biological variation???? did i get it right??
I'll be honest, I haven't thought a lot about this video in the past 7 years, however, if we just do biological replicates, than the average biological variation and technical variation both converge on 0, which is the goal. If we add technical replicates, then it takes longer (more replicates are needed) to converge on 0.
best video I have seen !! save me from struggle of seq data ! I am confused : should not the variation be the summed (difference)^2 ? do we assume the variations ~normal distribution ?
+Joshua Starmer Thanks! I think I got that , just for clarify : when I want only one number to characterize how spread the whole distribution does, I choose square the differences and sum them up to get that one number. But in this case , I just want to rewrite the read counts for each of the samples in the distribution in a more systematic way (mean + variance) .
Sure, you can do that - and people do it all the time when they don't have many cells to work with - but if you can get the cells, you're better off getting more biological samples.
I don't understand... Technical replicates should be nested in each biological replicate. Say you have technical replicated in triplicate or duplicate, the values should be fairly consistent between them, which is kind of the point of technical replicates, to control for stuff like pipetting error etc.. It seems dishonest NOT to do technical replicates, because who knows if what you're observing isn't just due to pipette error or undergraduate slave labor.
Exactly. This should be a bietapic process where you end up working with the technical average/median value and with some error value per measure. You cannot analyze technical and biological replicates at the very same time.
Hi, I have a doubt about this video. It is clear that biological replications are necessary, but I think that the technical replications are also essential. Imagine that you have the number of counts of the gene NANOG of mice 1 and it is 30, and you believe that because you don't have nothing to compare. Nevertheless, if you use 3 technical replications, and two of them have 120 and 113, and the third one is 30, you can expect that that 30 is more probable to be wrong. I think it is also important because it measure you also the external factors and batch effect, things very important in RNA-seq. ¿What do you think about? Thank you for all the videos :)
@Abhilash Kumar Tripathi Imagine you had a plate of bacteria and you used that to create a RNA-seq library. A technical replicate would consist of running that same library multiple times. If you grow up a new plate with the same species of bacteria and create a new RNA-seq library, then that is a biological replicate. Does that make sense?
Support StatQuest by buying my book The StatQuest Illustrated Guide to Machine Learning or a Study Guide or Merch!!! statquest.org/statquest-store/
As a master student I have to thank you for your videos. Without them, getting into the analysis of RNAseq-data would be a major chore. Thank you and stay healthy!
Thank you! :)
pretty sure this just saved my phd- triple bam!
Hey man, your videos about different RNA-seq topics and problems are more than awesome, I (and probably others) really appreciate this. Keep up the good work with this and your other StatQuest videos!
+Joshua Starmer Just one question about this video (I'm new to this): When you say 'read-counts', what exactly are you referring to? The numbers of reads of length k obtained from a gene X, during an RNA-seq experiment?
And that number is different for each individual (ignoring the technical problems) because of different gene expression, even if the individuals are genetically identical?
And also, is the combined length of all reads from a single gene X equal to the length of the whole gene X? I.e. the whole expressed gene is sequenced, or are some parts left out?
Sorry if I'm bothering, couldn't find a nice explanation and was hoping you could provide a little insight whenever you have time.
+Joshua Starmer This really cleared things up! Have a great day!!
great explanation, something RNA-seq world needs to come into terms with.
Thanks!
would you please make more videos? you are such a great educator and a master !! I highly recommend your video to my classmates in bioinformatics.
Please add explanatory video about stranded rna seq and their protocols in illumina because I could never understand the rna seq clustering and amplification step for that. Also about paired end reads using examples of sequences because its hard to understand the reverse complement and complement reads trhings.
Hiii. I would be glad to read a paper about this. Yesterday I was searching and I didn't find any paper talking about biological vs technical replicates in RNA-seq. If I have to make a decision, I would do it based on a paper
Noted.
Your videos are the best Joshua.
Hi Josh! Love your song🥲. If I passage the cells grow in one dish into 3 dishes, Let them grow or give them the same type of treatment. Harvest each RNA from one dish as one sample. Send that 3 RNA samples to seq. Is these biological sample or technical samples? Thank you !
That sounds like a technical replicate.
@@statquest Thank you!
Thanks. you mean that the biological variation cancels out the technical variation?? and that is why we should have biological variation???? did i get it right??
I'll be honest, I haven't thought a lot about this video in the past 7 years, however, if we just do biological replicates, than the average biological variation and technical variation both converge on 0, which is the goal. If we add technical replicates, then it takes longer (more replicates are needed) to converge on 0.
@@statquest alright i got it. thanks a lot
best video I have seen !! save me from struggle of seq data ! I am confused : should not the variation be the summed (difference)^2 ? do we assume the variations ~normal distribution ?
+Joshua Starmer Thanks! I think I got that , just for clarify : when I want only one number to characterize how spread the whole distribution does, I choose square the differences and sum them up to get that one number. But in this case , I just want to rewrite the read counts for each of the samples in the distribution in a more systematic way (mean + variance) .
What about pooling several technical replicates from one biological replicate together?
Sure, you can do that - and people do it all the time when they don't have many cells to work with - but if you can get the cells, you're better off getting more biological samples.
I don't understand... Technical replicates should be nested in each biological replicate. Say you have technical replicated in triplicate or duplicate, the values should be fairly consistent between them, which is kind of the point of technical replicates, to control for stuff like pipetting error etc.. It seems dishonest NOT to do technical replicates, because who knows if what you're observing isn't just due to pipette error or undergraduate slave labor.
Hey Joshua, Maybe Could you clearly explain difference between biological and technical replicates?
Exactly. This should be a bietapic process where you end up working with the technical average/median value and with some error value per measure. You cannot analyze technical and biological replicates at the very same time.
Hi, I have a doubt about this video.
It is clear that biological replications are necessary, but I think that the technical replications are also essential. Imagine that you have the number of counts of the gene NANOG of mice 1 and it is 30, and you believe that because you don't have nothing to compare. Nevertheless, if you use 3 technical replications, and two of them have 120 and 113, and the third one is 30, you can expect that that 30 is more probable to be wrong. I think it is also important because it measure you also the external factors and batch effect, things very important in RNA-seq.
¿What do you think about? Thank you for all the videos :)
All of those things you mentioned, technical artifacts, batch effects etc, can be measured with biological replicates.
What's an example of a biological replicate? What does that even mean?
I've got a StatQuest that explains the difference between technical and biological replicates: ua-cam.com/video/Exk0OoRG0PQ/v-deo.html
@Abhilash Kumar Tripathi Imagine you had a plate of bacteria and you used that to create a RNA-seq library. A technical replicate would consist of running that same library multiple times. If you grow up a new plate with the same species of bacteria and create a new RNA-seq library, then that is a biological replicate. Does that make sense?