10. Transcriptomics I

Which statements regarding the generation of (single-cell) RNA-Seq libraries for Illumina sequencers are correct?

1) The human Cell Atlas is an international project that was finished in 2017 -> wrong, started 2016 and still ongoing

2) The total RNA content per human cell is ~10ng -> wrong, it is rather ~10-30 pg so 1000x less than 10ng (and ~0,25 mio mRNA molecules per cell)

3) Early barcoding refers to the pooling of cDNA different samples before library generation and thereby increases the throughput of library generation -> correct

4) Current single-cell RNA-seq technologies have allowed to generate scRNA-seq datasets for up to 1000 cells -> wrong, more – biggest dataset that has been published with RNAseq is over 1 million cells (so 1000x more)

5) UMIs can be used to identify cDNA molecules that have been generated by amplification and not by different reverse transcription events -> correct

Which statements regarding quantitative RNA-seq analysis are correct?

1) Blat is a commonly used to map RNA-seq reads -> wrong, Blat is too slow; common are e.g. Bowtie, BWA, STAR, NextGenMap, GSNAP

2) Normalization allows to compare absolute expression levels among samples -> wrong, does not allow to compare absolute expression levels (which is nearly impossible among samples); one normalizes to the total number of reads; one assumes that there is the same amount of RNA molecule per sample or: under the assumption that most of the genes are not differently expressed

3) Read counts per transcript are more difficult to estimate than read counts per gene -> correct

4) RPKM is a measure of the expression normalized for transcript length and normalized for the total number of reads sequenced -> correct

5) Demultiplexing is the process in which reads are separated based on their barcodes that label different RNA-seq libraries -> correct

Which of the following statements regarding statistics are correct?

1) If the mean and standard deviation of a normal distribution is 0 and 1, respectively 95% of the data fall between -1.96 and + 1.96 -> correct

2) To half the standard error of the mean for a normal distribution, one needs to increase the sample size by two-fold -> wrong, 4x/ four-fold increase would be necessary because sem = s/squareroot(N)

3) The central limit theorem states that for many populations distributions that are not normally distributed, the distribution of sample means will become increasingly close to a normal distribution as the sample size increase -> correct

4) The standard deviation cannot be negative -> correct (because square root)

5) If a sample of n=100 is taken from a population whose standard deviation is equal to 10 then the standard error of the mean is expected to be 10 -> wrong, sem = s/squareroot(N) with s=10 and N=100 -> 10/10 = 1