Kun:LabNotes/inSituSeq: Difference between revisions

From ZhangLabWiki
Jump to navigation Jump to search
Line 65: Line 65:
     [[Media:probe2padlockFISSEQ_2012_0bp_stemCellGenes.txt|./probe2padlockFISSEQ_2012_0bp_stemCellGenes.pl]] > [[Media:Dec2012_0bp_gap_probes_stemCellGenes.txt|Dec2012_0bp_gap_probes_stemCellGenes.txt]]
     [[Media:probe2padlockFISSEQ_2012_0bp_stemCellGenes.txt|./probe2padlockFISSEQ_2012_0bp_stemCellGenes.pl]] > [[Media:Dec2012_0bp_gap_probes_stemCellGenes.txt|Dec2012_0bp_gap_probes_stemCellGenes.txt]]
     [[Media:probe2padlockFISSEQ_2012_20bp_stemCellGenes.txt|./probe2padlockFISSEQ_2012_20bp_stemCellGenes.pl]] > [[Media:Dec2012_20bp_gap_probes_stemCellGenes.txt|Dec2012_20bp_gap_probes_stemCellGenes.txt]]
     [[Media:probe2padlockFISSEQ_2012_20bp_stemCellGenes.txt|./probe2padlockFISSEQ_2012_20bp_stemCellGenes.pl]] > [[Media:Dec2012_20bp_gap_probes_stemCellGenes.txt|Dec2012_20bp_gap_probes_stemCellGenes.txt]]
    
    
   wc -l Dec2012_stemCellGene_*
   wc -l Dec2012_stemCellGene_*

Revision as of 18:51, 6 December 2012

RNA in situ sequencing

Responsibilities

  • Rui: in situ rolony generation
  • Matt: design and testing of decoding padlock probes
  • Ho Suk: instrumentation (fluidic devices), automation, image analysis

Method developments

Probe design

  • Strategy: Experimentally evaluate the capture efficiencies of 6,500 probes(or 13,000, depending on the minimal order allowed by Agilent), and pick the best ones.
  • Choice of genes:
  1. Variable expression among different brain regions and cell types (based on published ABI Brain Atlas data).
  2. Moderate abundance (too low: difficult to detect; too high: rolonies too crowded and difficult to resolve optically).
  3. Avoid alternatively spliced exons.
  4. Prioritizing genes used in ABI ISH mapping: Disease genes (383); Human cortical marker genes (167, 55 overlapped with disease genes).
  • Considerations:
  1. Targeting RNA versus cDNA (RNA: one fewer step, more specific, efficiency could depends on secondary structure; DNA: one more step, can be immobilized, secondary structure could be less a problem);
  2. Targeting nuclear transcripts versus cytoplasmic transcripts (space available for imaging, integration with single-cell toto-RNAseq data), nucleus versus cytoplasmic transcripts (based on the ENCODE RNAseq paper).
  3. Targeting 3'-ends or random positions (affects how cDNAs are generated, poly-T primed or N6/9 primed);
  4. Gap versus no gap (Specificity versus sensitivity);
  5. Targeting single exons or across splice junctions (mature transcripts versus nascent transcripts or gDNA);
  • Constrains:
  1. We can get only up to 240bp oligos from Agilent for now.
  2. A minimal list of features on the probes:
    1. Amplification adaptors (36-44bp, depending on the probe prep protocol).
    2. Common linker (48bp for our standard linker sequence, need to be reduced significantly, which is fine since we will be doing RCA instead of PCR and only a short annealing region is enough)
    3. Decoding region (6x20bp=120bp)
    4. Capturing arms (40-50bp, need to set an upper limit)
  • First round probe design.
    • A list of 5333 genes were generated by Dr. Yun Zhu from Wei Wang's group based on the following criteria:
      • Genes that show variable expression in the brain (based on PMID: 22996553).
      • Genes included in the ABI's ISH data.
      • Genes present in both cytosol and nucleus (based on the ENCODE RNAseq paper).
    • Matt selected up to 5 candidate regions per gene for probe design. One set of probes was designed with a gap size of 20bp. A second set has no gap.
    • I wrote a script to filter the probes to keep:
      • Up to 5 probes per gene.
      • All genes with names like "LOCxxxx" or "CxxOrfxxxx" are excluded because they are poorly annotated.
    • I then double-checked these probes by mapping to the human genome and reference mRNA sequences using NovoAlign. All probes that were not mapped to the refMrna, or were mapped to more than one location in the genome are excluded.
 ./probes2fasta.pl < outputFile_0gap_filtered.txt > outputFile_0gap_filtered.fasta
 ./probes2fasta.pl < outputFile_20gap_filtered.txt > outputFile_20gap_filtered.fasta
 /home/kunzhang/softwares/Novocraft/novocraft/novoalign -d /home/kunzhang/RNAseq/Data/CommonFiles/refMrna.ndx -f outputFile_0gap_filtered.fasta -F FA -r ALL > outputFile_0gap_filtered_novoalign_refMrna.out
 /home/kunzhang/softwares/Novocraft/novocraft/novoalign -d /home/kunzhang/softwares/Novocraft/novocraft/human_g1k_v37 -f outputFile_0gap_filtered.fasta -F FA -r ALL > outputFile_0gap_filtered_novoalign_human_g1k_v37.out &
 /home/kunzhang/softwares/Novocraft/novocraft/novoalign -d /home/kunzhang/RNAseq/Data/CommonFiles/refMrna.ndx -f outputFile_20gap_filtered.fasta -F FA -r ALL > outputFile_20gap_filtered_novoalign_refMrna.out
 /home/kunzhang/softwares/Novocraft/novocraft/novoalign -d /home/kunzhang/softwares/Novocraft/novocraft/human_g1k_v37 -f outputFile_20gap_filtered.fasta -F FA -r ALL > outputFile_20gap_filtered_novoalign_human_g1k_v37.out &
  • For 83 stem cell genes that no good probe was designed in the first pass, Matt repeat the probe design using mRNA sequences as the template. He generated two additional probe files, one for 20bp gaps, and one for 0bp gaps. I did the mapping to refMrna and human_g1k_v37.

Linker design

  • I wrote a script to generate a list of barcoded linker sequences. A total of 1792 linkers were generated using two colors and eight stages.
  • Since we will start with ~5000 candidate genes, more barcoded is needed. For this I increased the number of colors (fluorophores) to 3. With only 7 hybridization stages, we can decode 5040 genes. I also added another constrain that each barcode needs a minimal of two colors. The barcodes are also shorter, which gives more flexibility for the design of the common linker.

Probe assembly

  • The 0bp set used AP1V6/AP2V6 primers for amplification.
  • The 20bp set used AP1V4/AP2V4 primers for amplification.
   ./probe2padlockFISSEQ_2012_0bp.pl > Dec2012_0bp_gap_probes.txt
   ./probe2padlockFISSEQ_2012_20bp.pl > Dec2012_20bp_gap_probes.txt
   
   $ wc -l Dec*
   12344 Dec2012_0bp_gap_probes.txt
   12511 Dec2012_20bp_gap_probes.txt
   24855 total
  • I made some modification on the assembling scripts for the probes that were designed based on mRNA sequences. The reason being that they are multiple transcripts per genes, and some probes designed for different isoforms could be identical or overlapping.
   ./probe2padlockFISSEQ_2012_0bp_stemCellGenes.pl > Dec2012_0bp_gap_probes_stemCellGenes.txt
   ./probe2padlockFISSEQ_2012_20bp_stemCellGenes.pl > Dec2012_20bp_gap_probes_stemCellGenes.txt
  
  wc -l Dec2012_stemCellGene_*
  785 Dec2012_stemCellGene_0bp_gap_probes.txt
  843 Dec2012_stemCellGene_20bp_gap_probes.txt
  1628 total
  • I merged all assembled probe files. It turned out that there are still some redundant sequences in the first batch of brain gene probes. So I used the Excel "Remove Duplicates" function to remove all redundant sequences. This yielded a final list of 26,143 probes (228bp in length) that was submitted to Agilent for synthesis.

Rolonies generation

  • Considerations:
  1. Direct padlock capture versus whole transcriptome amplification

DNA in situ sequencing

Can test the probes on RNA/cDNA first to estimate the efficiency.