单细胞RNA-seq比对定量用什么工具好？使用哪个版本的基因组？数据来说话

这么多工具和基因组版本，选择困难症犯了，到底用哪个好呢？

2018 nature - Developmental diversification of cortical inhibitory interneurons : ENSEMBL release 84 Mus musculus genome

2017 Molecular Cell - Single-Cell Alternative Splicing Analysis with Expedition Reveals Splicing Dynamics during Neuron Differentiation : STAR, human genome (hg19), using GENCODE (v19) gene annotations; sailfish - GENCODE v19 protein-coding and long non-coding RNA annotation. Outrigger

2017 - Science - Single-cell RNA-seq reveals new types of human blood dendritic cells, monocytes, and progenitors : UCSC hg19 transcriptome; RSEM; TPM; 可行但是不完美，建议用count

2017 - Cell - Single-Cell Analysis of Human Pancreas Reveals Transcriptional Signatures of Aging and Somatic Mutation Patterns : cutadapt; hg19;

2015 - Cell Stem Cell - Single-Cell Transcriptome Analysis Reveals Dynamic Changes in lncRNA Expression during Reprogramming : TopHat; mm9; Cufflinks; DESeq

2017 - Nature - : UCSC mm10 mouse transcriptome using Bowtie; RSEM

小结：

QC: cutadaptb不错哦

如果只想进行定量，那就用bowtie、bowtie2比对，再用RSEM定量，这CNS用得最多；但是，单细胞能用TPM吗？显然不行，因为表达基因的数量差异太大了，这会带来很严重的偏差。

如果想要Reads count，那还是用FeatureCounts吧。（网上貌似说FeatureCounts比HTseq算法更好一些，但是HTseq2015年发表以来，引用了3000多次了，真是纠结选哪个！！！）

参考：Compariosn Htseq And Feature Count

http://bioinformatics.cvr.ac.uk/blog/featurecounts-or-htseq-count/

http://genomespot.blogspot.hk/2014/09/read-counting-with-featurecounts.html

如果想鉴定可变剪切，那就必须Tophat、Hisat2和STAR中选了，Hisat2引用少得可怜；为什么大家都不用呢？STAR的引用秒杀它，Tophat就太老了，不用也罢。