1. Presented By: Chinua Umoja
A Platform-Independent Method for Detecting Errors in Metagenomic Sequencing Data: DRISEE
Presented By:
Chinua Umoja

2. DRISEE
Designed primarily as a means to assess metagenomic sequencing error in samples submitted to MG-RAST.
(from website) an automated analysis platform for metagenomes providing quantitative insights into microbial populations based
The group also provides platform independent code that allows users to perform analyses on their own

4. Figure 2. DRISEE performance on simulated and real data.
Keegan KP, Trimble WL, Wilkening J, Wilke A, et al. (2012) A Platform-Independent Method for Detecting Errors in Metagenomic Sequencing Data: DRISEE. PLoS Comput Biol 8(6): e doi: /journal.pcbi

5. Figure 3. Total DRISEE errors of genomic and metagenomic data produced by 454 and Illumina technologies.

6. Figure 4. DRISEE error profiles for metagenomic sequencing data sets.

7. Figure 5. DRISEE calculated Errors, separated by error type, for 454 and Illumina metagenomic samples.

9. Step 1: Sequence Data
Considers data in FASTA format in attempt to keep DRISEE platform independent
Input consist of all sequenced reads produced in single run or sample

10. Step 2: Check for Random Sequencing
DRISEE was designed to consider sequences generated by random (shotgun) procedures.
a random fragment sequencing application that is used on a sample derived from a pool of organisms
Samples that exhibit nonrandom sequence patterns in their prefix are excluded

11. Step 3: Screen Reads
Sequence Data goes through a two step filtering procedure to remove two types of reads:
Reads that exhibit uncharacteristic lengths
Default if farther than two standard deviations from mean
Reads with ambiguous base calls

12. Step 4: Bin Reads
List of unique prefixes in the screened FASTA file is generated
Remaining reads are grouped according to their prefix

13. Step 5: Screen Bins Bins are sorted by read abundance
Bins with minimum number of reads are processed further

14. Step 6: Screen for Suspect Content
Bins can be screened for:
eukaryotic content
sequences with low complexity
known sequences that may exhibit an unusually high level of biological repetition

15. Step 7: End-Trim Reads
Reads within a bin are trimmed to uniform length

16. Step 8: Construct Consensus
All reads in a bin undergo MSA
currently uses QIIME
MSA are used to generate a consensus sequence

17. Step 9: Compare
Base-by-Base comparison is made with each individual read in a bin and final consensus sequence for that bin
Only the non-prefix portion of the read is compared
Read is scored at each position as either match or mismatch:
MATCH - A, T, C ,G insertion or deletion
MISMATCH- Asub, Tsub, Csub,Gsub, insertion or deletion

18. Step 10: Construct Bin-level DRISEE profiles
Deviations and matches for all reads in a bin are tallied with respect to position and the correlating consensus
Te consensus position indexed table of matches and mismatches for bin represents its bin-level DRISEE profile

19. Step 11: Construct run-level DRISEE Profiles
DRISSE profiles for all considered bins in a given run can be combined to produce DRISEE error profile for the run

20. Step 12: Sample Group Level DRISEE profile Construction
Combining DRISSE profiles for all bins in a group of runs
DRISEE profiles and the data they contain can be visualized directly or processed further to generate detailed analyses of information they contain