Skip to content

Latest commit

 

History

History
283 lines (202 loc) · 16 KB

Workflow_simplified.md

File metadata and controls

283 lines (202 loc) · 16 KB

Differential gene expression

The goal of this project was, starting from the sequencing files provided by the Sequencing Core of WTCHG, to generate a list of differentially expressed genes specific per cell line. This was also the first sequencing performed on HiSeq4000.

Experimental Design

Initially, five cell lines were planned to be analyzed.

  • Group 1: 4 CSF+ samples (sample 1, sample 6, sample 12, sample 16)

  • Group 2: 4 IL17A+ samples (sample 2, sample 7, sample 13, sample 17)

  • Group 3: 4 IFN+ samples (sample 3, sample 8, sample 14, sample 18)

  • Group 4: 3 IL17A+GMCSF+IFN- samples (sample 4, sample 9, sample 19)

  • Group 5: 3 CD45RA+Cyt- samples (sample 5, sample 10, sample 20)

  • Sample 11: these cells were sorted only on IFN+ by mistake and not IFN+, IL-17A-, GM-CSF-. Sample 11 will be most like group 3 but not as pure.

  • Sample 15: these cells are triple cytokine negative but not sorted on CD45. This sample will be most like group 5 but not as pure.

Genes that had to be expressed specifically for each cell type:

  • CSF2 should be expressed in GM-CSF cells (group 1, group 4)
  • IL17A should be expressed in IL17A+ cells (group 2, group 4)
  • IFNG should be expressed in IFN+ cells (group 3)
  • PTPRC should be expressed in all cells.

General overview of the sequencing data

This table contains general information (numbers of sequenced and mapped reads, number of reads retained after removing duplicates, number of reads mapped to exons, input concentration of RNA).

sample_name sequencing id input_yield sequenced reads mapped reads mapped reads nodup reads mapped to exons
CD45RA+Cyt- KNI1766A273 200 62,934,736 54,994,558 35,017,566 20,640,198
CD45RA+Cyt- KNI1766A288 200 68,135,862 59,552,750 35,526,346 21,086,445
CD45RA+Cyt- KNI1766A278 200 70,826,768 61,907,534 37,633,502 22,213,281
CD45RA_non_sorted KNI1766A283 200 68,602,461 59,885,416 39,839,932 23,385,406
CSF+ KNI1766A274 200 62,684,598 54,763,016 34,740,170 20,467,855
CSF+ KNI1766A280 200 67,455,873 59,183,454 34,195,690 20,201,799
CSF+ KNI1766A269 200 77,384,495 68,589,338 36,762,508 21,574,631
CSF+ KNI1766A284 76 70,221,073 61,069,882 34,349,526 20,535,774
IFN+ KNI1766A286 200 59,019,384 52,220,528 20,629,422 12,573,705
IFN+ KNI1766A271 200 70,522,903 62,017,526 38,567,234 22,550,540
IFN+ KNI1766A282 200 75,221,322 65,974,608 37,069,606 21,958,915
IFN+ KNI1766A276 76 61,592,829 54,616,810 25,000,494 14,887,863
IFN+_only KNI1766A279 200 69,670,672 61,124,116 34,200,938 20,189,447
IL17A+ KNI1766A275 106 62,310,684 55,454,766 15,185,274 9,584,108
IL17A+ KNI1766A270 200 78,393,511 68,992,988 34,702,600 20,666,285
IL17A+ KNI1766A281 35 47,909,489 42,749,000 9,637,426 6,341,500
IL17A+ KNI1766A285 41 33,478,324 29,799,156 6,193,486 4,138,692
IL17A+GMCSF+IFN- KNI1766A277 103 47,033,855 41,582,474 11,243,396 7,154,241
IL17A+GMCSF+IFN- KNI1766A272 200 58,098,730 51,781,252 32,145,574 18,511,109
IL17A+GMCSF+IFN- KNI1766A287 22 18,584,017 16,637,880 7,456,076 4,460,148

Plots with general statistics:
alt text alt text
alt text alt text

Expression of control genes

Samples with less than 6 mln mapped non-duplicate reads were not considered for this analysis. Outliar samples (sample 11 and sample 15) were not considered for this analysis either. Expression was calculated as read counts normalized for the number of reads mapped without duplicates.

  • CSF2, Ensembl ID ENSG00000164400.4 should be expressed in GM-CSF cells (CSF+ samples and IL17A+GMCSF+IFN- samples);
  • IL17A, Ensembl ID ENSG00000112115.5 should be expressed in IL17A+ cells (IL17A+ samples and IL17A+GMCSF+IFN- samples);
  • IFNG, Ensembl ID ENSG00000027697.8 should be expressed in IFN+ samples;
  • PTPRC, Ensembl ID ENSG00000081237.14 should be expressed in all cells.

Extra:

  • TBX21, Ensembl ID ENSG00000073861;
  • RORC, Ensemble ID ENSG00000143365.

alt text alt text alt text

Differential gene expression

Number of reads mapped to known genes was calculated with HTSeq on the sorted bam files without duplcates using GENCODE Ensembl annotation.

htseq-count --format=bam --stranded=reverse input.bam annotation

The number of reads mapped to known exons was relatively low -- around 60% of the number of mapped reads (after removing duplicates). However, differential expression analysis was still performed. The reason behind the low fraction of reads mapping to known exons remains unknown but is most probably due to a falty kit used to prepare total RNA for mRNA sequencing (Sequencing Core of WTCHG is investigating it at the moment).

  1. Removing ouliars (samples having low number of reads):
KNI1766A285	WTCHG_218577_217	Sample:17	IL17A+
KNI1766A287	WTCHG_218577_219	Sample:19	IL17A+GMCSF+IFN-

  1. Selecting genes with at least 100 reads mapped across all samples. This step is done to get rid of all zero-count lines, which will otherwise be considered as observations and will increase the FDR. This left us with 14,864 expressed genes.

  2. Running DESeq to get lists of differentially expressed genes.

Out of 14,864 expressed genes we have the following amount of differentially expressed genes.

FDR < 0.05

IL17A+GMCSF+IFN- IL17A+ IFN+ CSF+
CD45RA+Cyt- 1174 460 991 1311
CSF+ 10 2 53
IFN+ 203 42
IL17A+ 3

FDR < 0.1 (this will be used for the pathway analysis):

IL17A+GMCSF+IFN- IL17A+ IFN+ CSF+
CD45RA+Cyt- 1498 753 1348 1676
CSF+ 172 69 203
IFN+ 492 193
IL17A+ 71

Results of differentially expressed genes:

Files containing only genes with FDR < 0.05:

Response in pure cell populations

We would like to know e.g. which differences between the naive cells and the stimulated cells are consistent and which are unique per cell type.

Scroll down to the figure to see how we are comparing the DE between naive and stimulated cells from pure cell populations (CSF+, IFN+, IL17A+).

Lists of genes used to create the Venn diagram:

Double positive cell lines

Here we are checking whether double positive cell line (IL17A+GMCSF+INF-) is closer to CSF+, IFN+ or IL17A+.

These genes are differentially expressed only in IL17A+GMCSF+INF- against naive CD45+RA cells.

For this figure, lists of genes differentially expressed in double positives against naive cells and other, pure cell populations against naive cells.

Response against the naive Double positive cells
alt text alt text

Gene lists used in Venn diagrams:

CSF+, IL17A+ and IL17A+GMCSF+INF- against the naives

Here we are comparing differentially expressed genes between the naive cell line and any of the three following cell lines: CSF+, IL17A+ or IL17A+GMCSF+INF-.

scaled Venn Venn
alt text alt text

Note that variable sizes of DGE lists make such comparison biased (three times fewer genes were identified in IL17A+ compared to CSF+). This table contains both numbers and the percentages of DE genes in different categories. E.g., cell first left most cell 499 (38.03%) means that 499 genes are unique for CSF+, and 499 makes 38.03% of all genes identified as differentially expressed between CSF+ and naive. Row "common genes" contains the number of genes common for all three cell lines.

na CSF+ IL17A+ IL17A+GMCSF+INF-
CSF+ 499 (38.03%) 8 (1.74%) 390 (33.19%)
IL17A+ 8 (0.61%) 6 (1.3%) 30 (2.55%)
IL17A+GMCSF+INF- 390 (29.72%) 30 (6.51%) 340 (28.94%)
common genes 414 (31.55%) 414 (89.8%) 414 (35.23%)

Gene lists

To see the results of pathway analysis, click here.

Designed by Irina Pulyakhina [email protected]