General pre-processing of reads: All samples were sequenced by illumine Hiseq2000 with single end 100-bp read length. For libraries that generated from small RNA (PAR-CLIP and ribosome profiling), the adapters was trimmed by using FASTX-Toolkit34 (link). The deep sequencing data were mapped to Human genome version hg19 by Tophat version 2.035 (link) without any gaps and allowed for at most two mismatches. RIP and Ribosome profiling were analyzed by DESeq36 (link) to generate RPKM (reads per kilobase, per million reads). mRNA lifetime data were analyzed by Cuffdiff version 2.037 (link) to calculate RPKM.
Data analysis for each experiment: (1) for m6A profiling, the m6A-enriched regions in each m6A-IP sample were extracted by using the model-based analysis of ChIP-seq (MACS) peak-calling algorithm38 (link), with the corresponding m6A-Input sample serving as the input control. For each library, the enriched peaks with p < 1e-5 were used for further analysis; (2) for RIP, enrichment fold was calculated as log2(IP/input); (3) PAR-CLIP data were analyzed by PARalyzerv1.1 with default settings39 (link); (4) for ribosome profiling, only genes with RPKM>1 were used for analysis and the change fold was calculated as log2(siYTHDF2/siControl); (5) for mRNA lifetime profiling: RKPM were converted to attomole by linear-fitting of the RNA spike-in.
The degradation rate of RNA k was estimated by where t is transcription inhibition time (h), At and A0 represent mRNA quantity (attomole) at time t and time 0. Two k values were calculated: time 3 h versus time 0 h, and time 6 h versus time 0 h. The final lifetime was calculated by using the average of k3h and k6h.
Integrative data analysis and statistics: PAR-CLIP targets were defined as reproducible gene targets among three biological replicates (3,251). RIP targets (2528) were genes with log2(IP/input) >1. The overlap of PAR-CLIP and RIP targets were defined as CLIP+IP targets (1,277). And non-targets (3,905) should meet the conditions: (1) complementary set of PAR-CLIP targets; (2) RIP enrichment fold <0. For the comparison of PAR-CLIP and m6A peaks, at least 1 bp overlap was applied as the criteria of overlap peaks. Two biological replicates were conducted for ribosome profiling and mRNA lifetime profiling, respectively. And genes with sufficient expression level (RPKM>1) were subjected to further analysis. The change fold that used in the main text is the average of the two log2(siYTHDF2/siControl) values. Nonparametric Mann-Whitney U test (Wilcoxon rank-sum test, two sided, significance level = 0.05) was applied in ribosome profiling data analysis as previous reported22 (link). For the analysis of cell viability (Extended Data Fig.8e ), RPF of ribosome profiling data were analyzed by Cuffidff version 2.0 for differential expression test, and the genes that differentially expressed (p<0.05) were subjected to Ingeuity Pathway Analysis (IPA, Ingenuity System). RPF was chosen since it may better reflect the translation status of each gene.
Data accession: All the raw data and processed files have been deposited in the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo ). m6A profiling data are accessible under GSE 46705 (GSM1135030 and GSM1135031 are input samples while GSM1135032 and GSM1135033 are IP samples). All other data are accessible under GSE 49339.
Data analysis for each experiment: (1) for m6A profiling, the m6A-enriched regions in each m6A-IP sample were extracted by using the model-based analysis of ChIP-seq (MACS) peak-calling algorithm38 (link), with the corresponding m6A-Input sample serving as the input control. For each library, the enriched peaks with p < 1e-5 were used for further analysis; (2) for RIP, enrichment fold was calculated as log2(IP/input); (3) PAR-CLIP data were analyzed by PARalyzerv1.1 with default settings39 (link); (4) for ribosome profiling, only genes with RPKM>1 were used for analysis and the change fold was calculated as log2(siYTHDF2/siControl); (5) for mRNA lifetime profiling: RKPM were converted to attomole by linear-fitting of the RNA spike-in.
The degradation rate of RNA k was estimated by where t is transcription inhibition time (h), At and A0 represent mRNA quantity (attomole) at time t and time 0. Two k values were calculated: time 3 h versus time 0 h, and time 6 h versus time 0 h. The final lifetime was calculated by using the average of k3h and k6h.
Integrative data analysis and statistics: PAR-CLIP targets were defined as reproducible gene targets among three biological replicates (3,251). RIP targets (2528) were genes with log2(IP/input) >1. The overlap of PAR-CLIP and RIP targets were defined as CLIP+IP targets (1,277). And non-targets (3,905) should meet the conditions: (1) complementary set of PAR-CLIP targets; (2) RIP enrichment fold <0. For the comparison of PAR-CLIP and m6A peaks, at least 1 bp overlap was applied as the criteria of overlap peaks. Two biological replicates were conducted for ribosome profiling and mRNA lifetime profiling, respectively. And genes with sufficient expression level (RPKM>1) were subjected to further analysis. The change fold that used in the main text is the average of the two log2(siYTHDF2/siControl) values. Nonparametric Mann-Whitney U test (Wilcoxon rank-sum test, two sided, significance level = 0.05) was applied in ribosome profiling data analysis as previous reported22 (link). For the analysis of cell viability (
Data accession: All the raw data and processed files have been deposited in the Gene Expression Omnibus (
