Video

I got a similar error when using python 3.8. Switching to python 3.7 with cellbender solved this issue.

That is correct, good catch! Here's a nice video explaining the wet lab procedure in more detail: czcams.com/video/uFrrKHB9weY/video.htmlsi=F1fa6-XipaLcg1iq&t=406

The VirusDetect pipeline takes care of this internally. I don't remember the details, but this could be accomplished for example using the samtools command "samtools view -b -f 4 file.bam > unmapped.bam"

This is a live recording from a course, so unfortunately the sound quality is not the same as for the videos we make in a normal recording setup.

Aligning reads to reference and counting aligned reads per genes (e.g. with HTSeq) is done separate for each sample. You can then combine the count files together to a count table where the rows are genes and columns are samples. Using this table you can then look for statistically significantly differentially expressed genes with tools like DESeq2.

Good point! In this example data the mitochondrial genes had actually already been removed from the data. In part 2 it is shown how to filter out the spots that have high percentage of mitochondrial transcripts if your data has them.

This means that your file is corrupted. Please see the possible solutions proposed by Simon Andrews at github.com/s-andrews/FastQC/issues/37

For read quality we recommend the FastQC program, or MultiQC for many samples. We have integrated them (and many other programs) in the Chipster analysis software which runs in a Web browser. For more info, please see chipster.csc.fi/.

@@ChipsterTutorials Thank you and what if we zipped the FASTQ file , doest it be any problem? or it doesn't matter?

@@MissAsdfb99 gzipped FASTQ files work fine in Chipster, no problem :)

Thank you for your comment -please note, that this is a recorded live lecture of our guest lecturer. Thus we cannot redo this video.

@@ChipsterTutorials well then convey my feedback to your guest lecturer or next time select someone who does not stutter a lot

Maybe its due to fact that his primary language isn't English

You could do enrichment analysis for the lists of differentially expressed genes.

@@ChipsterTutorials Awesome! Thank you for your help, I really appreciate it!!

Or do we put the large amount (5-15) of PCs into t-SNE and UMAP to further reduce dimensionality until we are able to create one singular 2-D graph (2 dimensions)?

You got it right! Chipster (and the corresponding Seurat vignettes) give you few different plots for estimating the (true) dimensionality of the data, i.e. how many PCs to use for the next steps of the analysis. These plots usually show one or two components at once, and for example the heatmaps are plotted for first 12 PCs by default (you can tune this). I suppose it would be enough to show some of the plots to justify the choice for the number of PCs. So PCA is step 1 in reducing the dimensions, so that clustering step won't take for ever and struggle with the excess of information. Different plots showing the PCs are there to help you to choose the number of PCs you want to continue the analysis with: whether it's 10, or 15, or 50 first principal components. After clustering, tSNE and UMAP are used for visualisation: to really show the data in 2D (step 2 in dimension reduction).

@@ChipsterTutorials Thank you so much!! Your guides are so helpful for beginners like me :)

Of course you can, excellent questions! Those plots are from Chipster (chipster.csc.fi), but the codes within are pretty much directly from Seurat, so you can check the R-commands for example from here: satijalab.org/seurat/articles/pbmc3k_tutorial.html The heatmaps for the PCs show the "extreme" cells on the x-axis and "extreme" genes on y. They are "extreme" in their PCA scores, so those genes that basically best determine that particular principal component, i.e. the separation between the cells. Similarly for the cells: these cells "furthest away" (in the yellow or purple end) from each other on this spectrum of PC1. So what one might want to eye-ball with these plots is whether the genes reveal what that particular PC might be all about: for example, if the genes seem to be related to cell-cycle phase, one might want to consider regressing out that effect, or at least it's good to acknowledge this.

Great idea! Unfortunately, our knowledge on the topic is currently a bit limited. We'll keep this idea in mind!