Extended Data Fig. 10: RNA expression measurements using Smart-seq3 do not exhibit a systematic bias as a function of poly(A) tail length.
From: Single-cell quantification of ribosome occupancy in early mouse development
poly(A) tail length of transcripts from mouse GV-stage oocytes were retrieved from two previous studies a, Tail-Seq (n = 3)67 or b, PAIso-seq (n=2)68. Spearman rank correlation coefficients between our RNA-Seq measurements and poly(A) tail length was shown in the top left corners highlighting the lack of a strong relationship between the two variables for both Tail-Seq and PAIso-seq. c, Transcripts with shortest poly(A) tails (<35nt corresponding to the lowest 1% in TAIL-Seq and the lowest 3.7% in PAIso-seq) were compared to all other transcripts with respect to RNA expression measured by Smart-Seq3. The median of the distribution is shown with the horizontal line. The box depicts the interquartile range, the whiskers extend from the hinge to at most 1.5 times the interquartile range. d, Our RNA-seq measurements from zygotes using Smart-seq3 (n = 4), which uses oligo(dT) priming, was compared to measurements using SUPeR-Seq (n = 5)69, which provides poly(A) independent quantification. Spearman rank correlation between the two approaches is shown (p-value < 2.2 × 10−16). e, The log2 ratio of SUPeR-Seq to Smart-seq3 expression values (y-axis) were calculated for each gene and grouped by their respective poly(A) tail lengths (x-axis) as previously reported41. While there is not a systematic bias in expression measurements as a function of poly(A) tail length, transcripts with the shortest tails (<17nt) were slightly underestimated in Smart-seq3 (Kruskal-Wallis rank sum test; p-value 0.016).
