Sam:LabNotes/Microbiome-new/2011-4-14

From ZhangLabWiki
Revision as of 04:27, 27 April 2011 by >Sam Chiang (→‎Results)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Quantify low quality E.coli reads from Ecoli SAGs sequencing results[edit]

Background[edit]

  • Since bowtie mapping is very stringent, some E.coli reads (assuming they are low-quality E.coli reads) are not able to be mapped.
  • Total reads = mappled E.coli reads + non-mapped E.coli reads + non-mapped non-E.coli reads
    • The objective here is to quantify how many of those non-mapped E.coli reads.
  • We can't not directly BLASTN all of the failed-to-align reads since it will take extremely long time (> 1 week) using genemapster.

Possible approaches[edit]

Test1: BLATing all failed-to-align reads[edit]

  • BLAT failed-to-align reads to E.coli genome -> output in BLAST-output format -> MEGAN extraction of E.coli reads and get quantitative information.
  • Since the failed-to-aling reads (e.g. s_4_ID1.failed) are FASTQ file, I need to trasform them to FASTA files before I can BLAT them.
    • Using script fastq2fasta-hc.pl to transfer FASTQ format to FASTA format
BLAT command:
 
samchiang@genemapster:~/Tools/blatSuite34_64$ ./blat -out=blast /media/disk-1/samchiang/Raw-Read-Complete/101123_HL083/HC-MDA-Ecoli-SAGs-Ind1to8-SE/failed-or-nonspecific/s_4_ID1.failed.fa /media/disk-2/samchiang/Tools/bowtie-0.12.5/genomes/e_coli_k12_mg1655.fasta /media/disk-1/samchiang/Sub_tests_folder_imp/3_Quantify_lowquality_reads_Ecoli_SAGs/HL083_Ecoli_SAG_ID1.failed.fa.BLAT.txt &
  • Results
    • It is unclear why but The BLAT result (in BLAST output format) doesn't contain any header-string information. So MEGAN can't identify taxonomic information from BLAT results.
    • I did several tests and confirm that it's not due to the output format (blast or .psl) I chose. It's probably due to that the reference genome is a one string FASTA file.

Test2: bowtie mapping to gigantic reference genome[edit]

  • Whole genome de novo assembly by ABYSS -> BLASTN contigs -> MEGAN extraction of E.coli contigs -> Joint E.coli contigs with E.coli reference genome -> build up gigantic reference genome -> map the raw reads to gigantic reference genome to count the failed-to-align reads
  • Started from whole genome ABYSS-pe assembly 100up data (by Chris)
    • ID1_Ecoli.abyss.pe.K27.contigs.100up ~ ID8_Ecoli.abyss.pe.K27.contigs.100up
  • BLASTN the contigs and extract E.coli contigs by MEGAN
  • Cat the E.coli reference genome with the MEGAN extracted E.coli contigs
samchiang@genemapster:/media/disk-1/samchiang/Sub_tests_folder_imp/3_Quantify_lowquality_reads_Ecoli_SAGs/Step2_Megan_output/MEGAN output$ cat e_coli_k12_mg1655.fasta ID8_Ecoli_contigs_100up_including_summarized.new > Ecoli_gigentic_ID8 &
  • Build up bowtie gigantic E.coli genome
samchiang@genemapster:~/Tools/bowtie-0.12.5$ ./bowtie-build /media/disk-1/samchiang/Sub_tests_folder_imp/3_Quantify_lowquality_reads_Ecoli_SAGs/Step2_Megan_output/MEGAN_output/Ecoli_gigentic_ID2 indexes/Ecoli_gigentic_index_ID2/Ecoli_gigentic_index_ID2 &
  • Map the raw reads to Ecoli_gigantic_index
samchiang@genemapster:~/Scripts$ perl bowtie_gigantic_ID1_map /media/disk-1/samchiang/Raw-Read-Complete/101123_HL083/HC-MDA-Ecoli-SAGs-Ind1to8-PE/s_4_1and2_ID1.txt /media/disk-1/samchiang/Raw-Read-Complete/101123_HL083/HC-MDA-Ecoli-SAGs-Ind1to8-PE/s_4_1and2_ID1.failed.gig /media/disk-1/samchiang/Raw-Read-Complete/101123_HL083/HC-MDA-Ecoli-SAGs-Ind1to8-PE/s_4_1and2_ID1.nonunique.gig &

Results[edit]

File:04142011-quantifying low quality Ecoli reads-new3.png

Follow up[edit]

  • I did another similar test to make gigantic reference using E.coli reference + E.coli contigs (assembled from raw reads) + E.coli contigs (assembled from failed-to-align reads), the result wasn't improved either.

Discussion[edit]

  • The failed-to-align reads (from bowtie) were still failed to map against gigantic Ecoli reference genome.
  • Possible reason:
  1. There is a large portion of non-E.coli sequence in these failed-to-align reads, which prevent them from mapping to E.coli genome. Also bowtie mapping is too rigid on mapping algorithm.
  2. Those failed-to-align reads are hard to be assembled to contigs and maybe most of them are tossed during 100bp length filtering. The gigantic reference genome actually didn't include much "new template" for improving mapping.

Re-map regular Illumina library E.coli SAG using PE data[edit]

  • For the regular bowtie mapping, I re-do the whole raw read bowtie mapping using the pair-end reads, which has twice the data size than single-end reads I used for analysis on 12-08-'11.
  • I have higher (~10%) genome coverage rate since there are more input reads sequences.

Next[edit]

  • Directly BLASTn the failed-to-aligned reads using Triton cluster (suppress the reporting to 1 hit per read).