Following assembly in PASA, 23,898 out of the 29,294 Trinity contigs (81%) had valid ME49 genome alignments and were clustered into 16,462 transcripts (henceforth PASA transcripts), while 5,396 had invalid alignments and were excluded from PASA analysis (Additional files 1, 2, 3 and 4).
Invalid sequence (or indeed polymorphism) would affect the accuracy of our alignment calls and could have led to the inclusion of mismatched probes.
The large attitude error will make the linear alignment algorithm invalid.
However, trees based on concatenated alignments are invalid if the concatenated genes have conflicting histories (Leigh et al. 2008).
A final 511 base pair (bp) sequence (nucleotide position 1 to 511) was analysed, following trimming of invalid end sequences and sequence alignment.
Sensitivity (Sn) and specificity (Sp) were computed as follows: (3) (4) A sensitive classifier preserves as many valid sites as possible while a specific classifier minimizes the number of invalid sites in the final alignment.
In evaluating the performance of the classifier, it is important to remember that this type of data is not naturally balanced as there are usually more valid than invalid sites in a typical alignment.
Among those, 4472 were found to be invalid by the tests of alignment to Refseq RNA (Table 2: sum reverse complement + multiple genes), leaving 10847 reporters for further evaluation.
This is because most methods used to infer positive selection assume that the phylogenetic tree and branch lengths are constant across all sites in the alignment, a tenet that is invalid in the presence of recombination.
PASA aligned the Trinity assembled contigs to the ME49 genome using the Genome Mapping and Alignment Program (GMAP) [ 53] (gmap-2007-09-28), filthose invalid contigs (those most likely arising from sequencing and Trinity assembly errors), and reconstructed more complete transcripts from the Trinity contigs by reporting only the single best alignment.
