Matt:LabNotes/2015-3-19

From ZhangLabWiki
Jump to navigation Jump to search

CA12k_Nov2014_V4 in vitro Capture Sequencing Analysis[edit]

  • With in vitro Capture data, DARTFISH gene counts can be normalized and compared to other methods such as RNA-Seq
  • For Regression Analysis with RNA-Seq, FISSEQ, and BeadArray (PGP1f) see: Matt:LabNotes/2015-4-8#Regression_Analysis

Mapping MiSeq reads to Oligo Sequences[edit]

Convert Probelist to Fasta File[edit]

CA12kNov2014_Probelist2Fasta.pl

 bowtie2-build CA12k_Nov2014_V4_H1H2.fa CA12k_Nov2014_V4_H1H2
 bowtie2 --phred33 -x CA12k_Nov2014_V4_H1H2 -q MC-20150121_CA12kNov2014_V4gDNA-1.R1.fastq > CA12kNov2014_V4gDNA_R1_H1H2.sam 2> CA12kNov2014_V4gDNA_stderr.txt &
 2573875 reads; of these:
 2573875 (100.00%) were unpaired; of these:
   115745 (4.50%) aligned 0 times
   2458058 (95.50%) aligned exactly 1 time
   72 (0.00%) aligned >1 times
 95.50% overall alignment rate
 bowtie2 --phred33 -x CA12k_Nov2014_V4_H1H2 -q MC-20150121_CA12kNov2014_V4cDNA-2.R1.fastq > CA12kNov2014_V4cDNA_R1_H1H2.sam 2> CA12kNov2014_V4cDNA_stderr.txt &
 2425688 reads; of these:
 2425688 (100.00%) were unpaired; of these:
   110744 (4.57%) aligned 0 times
   2314903 (95.43%) aligned exactly 1 time
   41 (0.00%) aligned >1 times
 95.43% overall alignment rate
 samtools view -bS CA12kNov2014_V4gDNA_R1_H1H2.sam | samtools sort - CA12kNov2014_V4gDNA_R1_H1H2_sorted
 samtools view -h -F 4 CA12kNov2014_V4gDNA_R1_H1H2_sorted.bam > CA12kNov2014_V4gDNA_R1_H1H2_sorted_filtered.sam
 samtools view -bS CA12kNov2014_V4cDNA_R1_H1H2.sam | samtools sort - CA12kNov2014_V4cDNA_R1_H1H2_sorted
 samtools view -h -F 4 CA12kNov2014_V4cDNA_R1_H1H2_sorted.bam > CA12kNov2014_V4cDNA_R1_H1H2_sorted_filtered.sam

Count # of Reads Aligned to each Probe[edit]

CountReadsPer_Gene_Probe.pl

  • Verified with grep -c 'Gene' CA12kNov2014_V4gDNA_R1_H1H2_sorted_filtered.sam and grep -c 'Probe' CA12kNov2014_V4gDNA_R1_H1H2_sorted_filtered.sam
    • Differences in counts are accounted for by strings in header lines

Compare with Blue's BA8 bulk RNA-seq Gene Counts[edit]

RNA-Seq Data on Genome-miner[edit]

  • mapped_genes.fpkm_tracking.txt
    • /media/LTS_33T/RL_LTS33T/201404_201405_7Samples_BulkNucleiBatch1-20140623_Expt146/STAR/RL-BA8-sec9-n-N702-15May14_S2_mapped
      • neuronal nuclei
    • /media/LTS_33T/RL_LTS33T/201404_201405_7Samples_BulkNucleiBatch1-20140623_Expt146/STAR/RL-BA8-sec9-t-N701-15May14_S1_mapped
      • bulk tissue

Analysis in R[edit]

  • Some R commands used
  • Dropbox/GradZhangLab/CA12k_Nov2014/V4_CaptureAnalysis/V4CaptureAnalysis.RData
    • All_GeneCounts is data.frame with all data

Results[edit]

  • Since 66 (1.8%) probes have target regions that are highly repetitive (softmasked in reference genome; lowercase), they can have very high false in vitro capture counts and are removed from analysis
DARTFISH vs Sorted Neuronal Nuclei[edit]
  • Regression of only genes where DARTFISH does not target masked regions <-----------> Including genes targeting masked regions

File:BA8 logDARTFISH vs logRNASeqBulkN masked regression.jpeg File:BA8 logDARTFISH vs logRNASeqBulkN regression.jpeg

DARTFISH vs Bulk Tissue[edit]
  • Regression of only genes where DARTFISH does not target masked regions <-----------> Including genes targeting masked regions

File:BA8 logDARTFISH vs logRNASeqBulkT masked regression.jpeg File:BA8 logDARTFISH vs logRNASeqBulkT regression.jpeg

  • Regression of only genes where DARTFISH does not target masked regions
    • Excluded high and low expression genes based on RNA-Seq (-1 < log_FPKM < 8)

File:BA8 logDARTFISH vs logRNASeqBulkT masked medExpressGenes regression.jpeg

Conclusion[edit]

  • in situ DARTFISH gene counts correlates better with bulk tissue RNA-Seq
  • While removing genes with probes that target masked regions is conservative and certainly correct, it barely improves correlation with RNA-Seq
    • Below are histograms of invitro gene counts, LEFT: all probes for gene considered, MIDDLE: only genes with no probes targeting masked region, RIGHT: partially masked - all 240 genes included but probes that target masked region not counted

File:Hist CA12kNov2014 V4 ReadsPerGene allprobes.jpeg File:Hist CA12kNov2014 V4 ReadsPerGene masked.jpeg File:Hist CA12kNov2014 V4 ReadsPerGene partmasked.jpeg

  • Need to design 66 suppressor oligos to completely suppress (ordered probes flagged with a 1 in last column of file)
    • Maybe only 12 probes with >1,000 count to suppress the probes that have astronomical in situ counts
      • 18 probes with > 500, 22 probes with > 200, 32 probes with > 100
      • Use histogram of reads per probe (with softmaskedprobes removed) to decide

File:Hist CA12kNov2014 V4 ReadsPerProbe masked.jpeg File:Hist CA12kNov2014 V4 ReadsPerProbe maskedtargets.jpeg

  • Ordered 39 probes
    • 13 probes with >850 count
    • 26 probes (with 50 < count < 850 and >~30% lowercase) + (probes with 11 < count < 50 and >~30% lowercase and show up in DARTFISH data)